AI model finder · verified source catalogs
Find an AI model that fits your work
Search in plain English, then refine the full catalog with advanced filters.
AI-powered model search
Describe what you need
Detection runs in your browser. Searches and the selected hardware profile are logged to improve recommendations; no email address is collected here. Privacy details
Model results
2,069 matches · page 1 of 99 · 21 per page
Qwen: Qwen3 Next 80B A3B Thinking
Qwen: Qwen3 Next 80B A3B Thinking is a text-only reasoning model intended for difficult, multi-step tasks such as mathematical proofs, code synthesis and debugging, logic, and agentic workflows. The supplied OpenRouter description says it outputs structured “thinking” traces by default and identifies the model as reasoning-first. Tool-calling is listed as a capability, making it relevant for applications that combine deliberation with external actions or tools. The advertised context window is 262,144 tokens, which may support large prompts and extended reasoning workflows, although runtime limits and behavior are not specified in the supplied material. OpenRouter lists pricing of $0.15 per million input tokens and $1.20 per million output tokens. The record gives a parameter count of 80,000,000,000 for this model; it does not identify a quantized upload, so quantization details are unavailable. A model task classification and license are also not available in the supplied material, so licensing should be verified before deployment. No inference-speed benchmark is provided, and performance rankings or evaluation results are not available here.
- Model size
- 80B parameters
- Context
- 262,144
- Knowledge cutoff
- 2025-09-30
- Input price
- $0.15/M
Solade-7B-4
Solade-7B-4 is listed as a text model with 7 billion parameters and a reasoning use case, but the supplied Hugging Face excerpt does not verify this exact repository. The excerpt instead describes DLMveloper/Solade-Qwen-8B-4, a text-generation model with 8B parameters, so its metadata should not be applied to Solade-7B-4. Based on the model record, Solade-7B-4 accepts text input and is intended for reasoning-oriented workloads. No specific capabilities, context window, quantization format, license, inference-speed benchmark, pricing, architecture, training data, or runtime requirements are available in the supplied material. The record also does not identify a more precise model task beyond the text modality, and it does not establish whether this upload is a quantized variant or a base model. Developers should therefore treat the 7-billion-parameter figure and reasoning categorization as the available record-level metadata, while verifying the repository contents, model card, license terms, supported inference stack, and performance characteristics before deployment.
- Model size
- 7B parameters
- Context
- Not listed
MoonshotAI: Kimi K3.1 Pro (batch)
MoonshotAI: Kimi K3.1 Pro (batch) is a multimodal reasoning model available through OpenRouter’s batch endpoint for developers working with long inputs and mixed media. The supplied material describes it as Moonshot AI’s flagship open-weight model and a post-trained successor to Kimi K3, with an advertised 1,048,576-token context window. Its listed task is image-text-to-text, and its modalities are text, image, and video. Practical use cases include reasoning, vision-based workflows, video understanding, and long-horizon coding; the supplied summary also references 3D work, though no further 3D-specific details are provided. The parameter count recorded for this exact model is 2,780,110,363,392, while no inference speed benchmark is available in the supplied material. Quantization is unspecified, so this record does not establish that the endpoint is a quantized upload or provide quantization details. The license is recorded as “other”; the applicable licensing terms should therefore be reviewed before deployment. The material does not provide benchmark results, safety classifications, or runtime defaults beyond the advertised context window.
- Model size
- 2.8T parameters
- Context
- 1,048,576
- Input price
- $0.7/M
Thinking Machines: Inkling 2 Small
Thinking Machines: Inkling 2 Small is an open-weight multimodal mixture-of-experts reasoning model intended for image-and-text-to-text tasks. It accepts text and image inputs, supports tool-calling, and is positioned for reasoning and vision workloads. The advertised context window is 1,048,576 tokens, which may suit long documents or extended multimodal interactions, although the supplied material does not specify a runtime default or deployment configuration. The record lists Apache-2.0 licensing for this model and does not specify quantization, so this listing should be treated as the base model rather than a documented quantized upload. Parameter information needs careful interpretation: the supplied record lists 383,072,987,216 parameters, while the source description characterizes the model as having 17B active parameters out of 594B total, reflecting the distinction between active and total mixture-of-experts capacity. OpenRouter lists pricing of $0.30 per million input tokens and $1 per million output tokens. No inference tokens-per-second benchmark is available in the supplied material. The model may be a fit for multimodal reasoning and tool-assisted applications, but implementation details such as supported tools, image constraints, and benchmark results are not provided.
- Model size
- 383.1B parameters
- Context
- 1,048,576
- Input price
- $0.3/M
Thinking Machines: Inkling 2 Small (free)
Thinking Machines: Inkling 2 Small (free) is an open-weight multimodal mixture-of-experts reasoning model for image-text-to-text tasks. The supplied OpenRouter material describes it as a Thinking Machines Lab model with 17B active parameters out of 594B total, while this record lists a parameter count of 383,072,987,216; those figures are not consistent, so the active-versus-total description should be treated separately from the recorded count. It accepts text and images, supports tool-calling, and is positioned for reasoning and vision workloads. The advertised context window is 1,048,576 tokens, which is a model capability rather than a guaranteed runtime configuration. The OpenRouter variant is marked free, with input and output pricing recorded as zero. The license for this record is Apache-2.0. Quantization is listed as null: the supplied material does not identify a quantized upload, so quantization details remain unverified. No inference speed or tokens-per-second benchmark is provided. Developers should also note that the supplied excerpt does not provide benchmark results, deployment requirements, safety characteristics, or detailed tool-use constraints.
- Model size
- 383.1B parameters
- Context
- 1,048,576
- Input price
- $0/M
neuroblast-v4.1-450M
neuroblast-v4.1-450M is a 450-million-parameter Hugging Face model intended for reasoning, but the supplied material provides limited technical detail for evaluating it. The source page describes it as a machine-learning model with approximately 0.4B parameters, consistent with the record’s parameter count of 450,000,000. The record identifies a 32-bit format; this indicates the stored numeric precision, but the supplied material does not establish whether this is a quantized derivative of another model or the base model itself. No specific model task, advertised context window, inference speed, benchmark results, pricing, or runtime requirements are available in the supplied material. Modalities and additional capabilities are also unspecified, so its practical use is limited to the recorded reasoning use case rather than verified coding, tool-calling, vision, or other functions. The model card is hosted at Hugging Face under mkurman/neuroblast-v4.1-450M. A license is not available in the supplied record, so redistribution and commercial-use terms remain unverified for this specific model. Developers should therefore treat the parameter count, 32-bit format, and reasoning categorization as the principal known attributes, with compatibility and performance requiring separate validation.
- Model size
- 450M parameters
- Context
- Not listed
Ling-3.0-tiny-Uncensored-Abliterated-i1-GGUF
Ling-3.0-tiny-Uncensored-Abliterated-i1-GGUF is an 8-billion-parameter model repository on Hugging Face, intended for local use through a GGUF-formatted upload. The exact quantization level is not available in the supplied material, so hardware requirements, memory usage, and expected runtime performance cannot be verified here. Its name explicitly identifies it as uncensored and abliterated, and the record lists cybersecurity and reasoning among its practical use cases. No supported modality is specified, and the model task is not provided, so this listing should not be treated as confirming vision, audio, image, or video capabilities. The source excerpt does not provide a context-window value, inference benchmark, pricing, or other performance measurements. A license is also not specified for this record; review the repository before redistribution or commercial deployment. This upload should be distinguished from any underlying Ling base model because the supplied record identifies this repository specifically as an i1 GGUF variant, while providing no verified details about the conversion method or the relationship to the base model.
- Model size
- Size not listed
- Context
- Not listed
VideoGuard-Qwen3.5-9B-Safety-RL-Uncensored
VideoGuard-Qwen3.5-9B-Safety-RL-Uncensored is a 9B-parameter Hugging Face model presented as a video-text-to-text model. Its name explicitly identifies it as uncensored, while the supplied record lists image and text modalities and associates it with vision, video, and reasoning use cases. This makes it potentially relevant for developers exploring multimodal analysis, video-oriented prompting, or reasoning workflows, but the available material does not verify the exact video input interface, supported formats, or runtime integration requirements. The record does not identify a quantization format, so this should not be treated as a quantized upload without further evidence. A license is not available in the supplied material, which is a material consideration for redistribution or commercial deployment. Context-window size, inference speed, pricing, benchmark results, and hardware requirements are also unavailable. The source page describes the model category and parameter size but provides no additional technical specifications in the supplied excerpt. Developers should therefore treat its advertised video-text-to-text positioning and uncensored classification as the primary verified attributes, with implementation and usage constraints still requiring confirmation from the model files or documentation.
- Model size
- 9B parameters
- Context
- Not listed
wizard-math:13b
wizard-math:13b is a 13-billion-parameter model intended for mathematical and logical reasoning. It may be a useful choice for developers building applications that need help solving math problems, working through logic questions, or producing reasoning-oriented answers. The supplied material identifies reasoning as its use case, but does not specify additional capabilities such as coding, tool calling, vision, audio, or image generation; no input or output modalities are listed. An advertised context window is not available in the supplied material, so the amount of conversation or problem context it can reliably process cannot be evaluated here. The source also provides no inference-speed benchmark, pricing information, or licensing terms. Developers should therefore validate response quality, latency, memory requirements, and redistribution or commercial-use conditions before deployment. The model is listed through Ollama under the exact name wizard-math:13b, and the source page describes it simply as focused on math and logic problems. Its suitability for broader general-purpose workloads, complex multi-step reliability, and production guarantees is not established by the supplied information.
- Model size
- 13B parameters
- Context
- Not listed
wizard-math:7b
wizard-math:7b is a 7-billion-parameter model intended for mathematical and logic-focused reasoning. It may be a suitable candidate for developers building applications that need assistance with equation solving, structured logical analysis, or other math-oriented problem-solving workflows. The supplied material identifies reasoning as its use case, but does not provide narrower task categories, benchmark results, accuracy measurements, latency data, or an inference-speed report. No modalities or additional capabilities are listed, so support for vision, audio, image generation, tool calling, or other input and output formats is not established here. The advertised context window is not available in the supplied material, and no runtime context configuration is provided. Pricing information is also unavailable. The source page is hosted by Ollama and shows an explicit update date of December 19, 2023; verify whether the currently served artifact matches your deployment requirements. Licensing terms are not available in the supplied material, so review the applicable model and distribution terms before commercial or redistributed use. Overall, consider it for math and logic reasoning experiments, while validating quality, context needs, operational performance, and licensing independently.
- Model size
- 7B parameters
- Context
- Not listed
wizard-math:70b
wizard-math:70b is a 70-billion-parameter model presented as focused on math and logic problems. Its clearest supported use case is reasoning, particularly tasks involving mathematical problem solving, formal logic, and related analytical workflows. The supplied material does not list specific input or output modalities, additional capabilities, a context-window size, pricing, or an inference-speed benchmark, so these details are unavailable for evaluating production integration. No runtime configuration or default context setting is provided either. The model page identifies the version as wizard-math:70b and shows an explicit update date of December 19, 2023; this is the only source-freshness detail available. Developers should therefore validate current availability, quality, latency, memory requirements, and compatibility in their own Ollama environment before deployment. Licensing terms are not included in the supplied material, so review the applicable model and distribution license separately before using it commercially or redistributing it. Based on the available description, this model is best considered for focused mathematical and logical reasoning rather than general multimodal, coding, audio, image, or tool-use workloads.
- Model size
- 70B parameters
- Context
- Not listed
orca2:7b
orca2:7b is a 7-billion-parameter model intended primarily for reasoning and research. It was built by Microsoft Research as a fine-tuned version of Meta’s Llama 2 models, so it may be worth evaluating for workloads that prioritize reasoning behavior over specialized capabilities. The supplied model record lists reasoning and research as use cases, but does not identify coding, tool-calling, vision, audio, image-generation, or other modality-specific capabilities. It also provides no advertised context-window size, inference-speed benchmark, or pricing information, so those factors should be measured or configured based on your deployment environment rather than assumed. The source page describes the model as designed to excel particularly in reasoning, but supplies no comparative benchmark results or reliability guarantees. Licensing terms, acceptable-use conditions, and deployment restrictions are not available in the supplied material and should be verified before production use. The page is explicitly dated November 22, 2023, so its descriptive information may warrant additional validation before adoption.
- Model size
- 7B parameters
- Context
- Not listed
orca2:13b
orca2:13b is a 13-billion-parameter model available through Ollama, intended primarily for reasoning and research workflows. The supplied description says it was built by Microsoft Research as a fine-tuned version of Meta’s Llama 2 models, with particular emphasis on reasoning. This makes it a candidate for developers exploring analytical or research-oriented language-model tasks, especially when running the model through the Ollama ecosystem. However, the supplied material does not provide benchmark results, an inference-speed measurement, pricing, or an advertised context-window size, so its throughput, operating cost, and long-context suitability cannot be assessed here. No modalities or additional capabilities are listed, meaning image, audio, video, tool-calling, and other specialized support are not established by the available information. The source excerpt also does not include licensing terms, safety guidance, hardware requirements, quantization details, or deployment recommendations; these should be verified before production use. The model page identifies the underlying family as Llama 2 fine-tuned by Microsoft Research, but it does not provide further evaluation data for this specific 13B variant.
- Model size
- 13B parameters
- Context
- Not listed
internlm2:7b
internlm2:7b is a 7B-parameter model presented for practical scenarios, with the source specifically advertising outstanding reasoning capability. It may be worth evaluating for reasoning-oriented applications where a relatively compact model is useful, including analysis, structured problem solving, and other developer workflows that can tolerate validating outputs. The supplied material does not specify supported modalities or additional capabilities such as vision, audio, tool calling, coding, or agentic behavior, so those should not be assumed. It also does not provide a context-window size, inference-speed benchmark, pricing, or runtime configuration; these details are unavailable in the supplied material. The model is distributed through Ollama under the `internlm2:7b` identifier, which may simplify local evaluation and deployment through that ecosystem. Before production use, verify the model card and applicable license terms, because licensing information is not available in the supplied material. The description identifies the model as InternLM2.5, while the Ollama page uses the internlm2:7b repository name; confirm that this tag matches the variant you intend to run.
- Model size
- 7B parameters
- Context
- Not listed
wizardlm2:7b
wizardlm2:7b is a 7-billion-parameter, text-only language model intended for developers building conversational, reasoning, multilingual, and agent-oriented applications. The supplied model page describes it as a Microsoft AI model with improved performance for complex chat, multilingual tasks, reasoning, and agent use cases. Its listed practical uses are agents and reasoning, making it a candidate for text-based assistants, workflow automation, and applications that need multi-step language understanding. The supplied material does not specify tool-calling support, structured or typed outputs, coding performance, vision, audio, or other non-text modalities, so those capabilities should not be assumed. A context-window size and inference-speed benchmark are not available in the supplied material, which limits direct comparison for long-document workloads or latency-sensitive deployments. The page shows an explicit update timestamp of April 16, 2024. Licensing terms are not available in the supplied material and should be checked before production or redistribution. The model is available through Ollama at the provided official model URL.
- Model size
- 7B parameters
- Context
- Not listed
internlm2:1m
internlm2:1m is presented as an InternLM2.5 model with 7B parameters, intended for practical scenarios and advertised with outstanding reasoning capability. Its clearest supported use case is reasoning, such as analytical question answering or other tasks requiring multi-step thought. The supplied material does not list additional capabilities, modalities, tool use, coding support, or agent features, so developers should not assume those functions without further validation. No context-window size, inference-speed benchmark, input or output pricing, or licensing information is available in the supplied material. The page identifies the model as updated on August 16, 2024, which provides a dated reference point but does not establish current performance or maintenance status. The `:1m` variant name and the page's 7B description should be checked against the intended Ollama package before deployment, especially if storage, memory, or runtime requirements matter. Overall, this model is most reasonably evaluated for general reasoning workloads, while important operational, compatibility, and licensing details require confirmation from additional documentation.
- Model size
- 1M parameters
- Context
- Not listed
phi:2.7b
phi:2.7b is a compact text language model suited to lightweight reasoning and research-oriented applications. The supplied material identifies it as Phi-2, a 2.7B model from Microsoft Research, and advertises reasoning and language-understanding capabilities. Its listed modality is text, so it is appropriate for text-based analysis, question answering, summarization, and exploratory research workflows; reasoning and research are the supported use-case classifications. The model may be attractive when a smaller parameter footprint is more important than access to multimodal input, but the supplied material does not provide benchmark results, inference speed, a context-window specification, or pricing. It also does not provide licensing terms, so licensing suitability should be verified before deployment or redistribution. The Ollama page identifies the model as updated on January 6, 2024; this is the only explicit freshness date in the supplied material. Runtime behavior, memory requirements, and quality on a particular workload are not available here and should be evaluated in the intended Ollama environment.
- Model size
- 2.7B parameters
- Context
- Not listed
internlm2:1.8b
internlm2:1.8b is a compact InternLM2-family model positioned for reasoning and practical scenarios. The supplied Ollama page describes InternLM2.5 as having outstanding reasoning capability, making this model potentially relevant for lightweight reasoning-oriented applications. However, the page summary refers to a 7B model while the exact model name is internlm2:1.8b and the supplied record lists 1.8 billion parameters; this inconsistency should be resolved before making capacity or hardware decisions. No context-window size, inference-speed benchmark, modalities, or additional capabilities are available in the supplied material. Consequently, the material supports reasoning as the primary use case, but does not establish suitability for coding, vision, audio, tool calling, agents, or other specialized workflows. Pricing information is also unavailable. The page identifies the model through Ollama, but licensing terms are not provided in the supplied excerpt, so review the applicable model and distribution licenses before deployment. The page includes an explicit update date of August 16, 2024, although no newer model-specific technical details are supplied here.
- Model size
- 1.8B parameters
- Context
- Not listed
internlm2:20b
internlm2:20b is an Ollama-listed language model intended primarily for reasoning-oriented workloads. The directory record identifies it as having 20 billion parameters, while the supplied Ollama summary describes “InternLM2.5” as a 7B parameter model; this inconsistency should be resolved before relying on the advertised size for hardware planning. The source characterizes the model as tailored to practical scenarios and highlights reasoning capability, making it a candidate for general analytical or problem-solving tasks where reasoning is the main requirement. No coding, tool-calling, agentic, vision, image-generation, audio, or other specialized capabilities are explicitly documented in the supplied material. Modalities are not specified, and an advertised context window is unavailable, so prompt-length assumptions should be verified in the runtime or model documentation. No inference speed benchmark, input or output pricing, or licensing information is provided in the supplied material. The page includes an explicit August 16, 2024 update timestamp, so its metadata may require verification before deployment. Consider testing it against representative reasoning tasks and confirming the actual checkpoint, context limits, quantization, and license in your intended Ollama setup.
- Model size
- 20B parameters
- Context
- Not listed
mathstral:7b
mathstral:7b is a 7-billion-parameter model intended primarily for mathematical reasoning and scientific discovery. It is distributed through Ollama and is attributed in the supplied description to Mistral AI. This makes it a candidate for tasks such as solving mathematical problems, working through formal reasoning steps, exploring scientific questions, and supporting research-oriented workflows. The available material does not identify additional capabilities, supported modalities, tool-calling behavior, coding specialization, or structured-output guarantees, so those areas should be validated before integration. A context-window size is not available in the supplied material, and no inference-speed benchmark or pricing information is provided. The model page identifies the release as a 7B variant, but does not provide further architectural or quantization details in the available excerpt. Licensing terms are also not available in the supplied material; review the applicable model and Ollama distribution terms before deploying it commercially or redistributing it. Overall, it is best evaluated as a focused local model for mathematics and science rather than assumed to be a general-purpose assistant.
- Model size
- 7B parameters
- Context
- Not listed
codegemma:2b
For developers, codegemma:2b is a lightweight model aimed primarily at practical coding and reasoning tasks. The supplied description lists fill-in-the-middle code completion, code generation, natural-language understanding, mathematical reasoning, and instruction following, making it potentially useful for completing code within an editor, generating code from requirements, explaining or interpreting developer requests, and handling programming-related reasoning. Its name indicates a 2-billion-parameter variant, which may be relevant when evaluating local deployment constraints, although no hardware requirements or performance benchmark is provided here. The supplied material does not identify supported modalities beyond the listed text-oriented capabilities, and it does not provide an advertised context-window size, pricing, inference speed, or runtime configuration. Licensing terms are not available in the supplied material, so review the applicable model and Ollama terms before redistribution or commercial use. The description presents CodeGemma as a collection rather than detailing this variant’s task-specific quality, so validate completion accuracy, instruction adherence, and mathematical reasoning on your own workloads before relying on it in production.
- Model size
- 2B parameters
- Context
- Not listed