Language Models

Bielik 1.5B v3 Instruct is a Polish language model specialized in generating text responses based on system instructions and user prompts, optimized for conversational tasks.

Cordia-A6 Text Generation Model generates text sequences from prompts using configurable parameters for length, randomness, and repetition control.

Claude Sonnet 4 is a hybrid reasoning model providing near-instant responses and extended thinking capabilities for coding, analysis, and complex instruction following.

OpenAI O1-Mini Language Model is a compact, cost-efficient language model built for real-time, high-throughput applications like chatbots, coding tools, and document processing with strong reasoning and formatting capabilities.

o1 retired from OpenAI’s API on October 23, 2026. Check the exact shutdown notice, historical model details and GPT-5.6 Sol migration guidance.

GPT-4o is deprecated by OpenAI, with listed snapshots due to shut down January 14, 2027. Check its text/image contract and Replicate access separately.

OpenAI deprecated GPT-4o mini on October 14, 2026, with listed API snapshots due to shut down January 14, 2027. Verify Replicate access separately.

Meta Llama 3 8B is a large language model optimized for text generation tasks, accepting prompts and controlling output length through token limits.

Fast Qwen3 32B generates text responses based on input prompts using an optimized language model with optional intermediate reasoning.

o4-mini retired from OpenAI’s API on October 23, 2026. Check the exact shutdown notice, historical model details and GPT-5.6 Terra migration guidance.

GPT-4.1 nano retired from OpenAI’s API on October 23, 2026. Check the exact shutdown notice, historical model details and GPT-5.6 Luna migration guidance.

OpenAI deprecated GPT-4.1 mini on October 14, 2026, with listed API snapshots due to shut down January 14, 2027. Verify Replicate access separately.

OpenAI deprecated GPT-4.1 on October 14, 2026, with listed API snapshots due to shut down January 14, 2027. Verify Replicate access separately.

Kimi-VL-A3B-Thinking is a multimodal large language model specialized in complex reasoning tasks that processes images and text to generate detailed responses with explicit thinking processes.

Nucleum Nano - Distilled Language Model is a distilled language model designed for text generation and deep analytical reasoning tasks with configurable parameters.

Llama 4 Maverick Instruct is a mixture-of-experts language model for generating text responses to prompts.

Meta Llama 4 Scout Instruct is a multimodal-capable language model using mixture-of-experts architecture for text generation and instruction-based tasks.

Qwen2.5-Omni Multimodal AI is a multimodal model that processes text, images, audio, and video inputs to generate text responses and natural speech, supporting real-time streaming interactions for conversational applications.

DeepSeek V4 Pro 0813 current API contract and pricing, with the remaining conflict in an older redirect announcement.

Claude Fable 5 is Anthropic's June 2026 long-horizon model and the predecessor to Fable 5.1. Check its API specs, pricing, and upgrade boundary.

Review Claude Opus 4.8 as the active predecessor to Opus 5, including adaptive thinking, 1M context, pricing, and migration use cases.

Claude Mythos 5 is Anthropic's June 2026 restricted-access model and predecessor to Mythos 5.1. Check its Project Glasswing access and specs.

Review Holo3.1 35B-A3B model details: H Company API ID, text and image input, 65,536 context, function calling, pricing, free tier, and local checkpoints.

Explore GLM 5.2 from Z.AI with supported features, long-context coding use cases, function calling, structured output, MCP, pricing signals, and community usage patterns.

Review GPT-5.6 Sol's temporary $4 input and $20 output API pricing, 1.05M context, 128K output limit, cache rates, and long-context pricing.

Review GPT-5.6 Terra API pricing, 1.05M context, 128K output limit, cached-input rates, long-context pricing, and balanced positioning.

Review GPT-5.6 Luna's reduced API pricing: $0.20 input, $0.02 cached input, and $1.20 output per 1M tokens, plus context and limits.

Review Claude Sonnet 5 details: Anthropic model ID, 1M context, 128K max output, coding and agent use cases, platform availability, and token pricing.

Review Grok 4.5 details: xAI model ID, Cursor availability, coding and agent use cases, API pricing, EU rollout caveat, and model-routing notes.

Review Kimi K3 details: 2.8T sparse MoE architecture, 1M context, native vision, API pricing, official open weights, license, and deployment.

Review Gemini 3.6 Flash API pricing, 1M context, 64K output, multimodal inputs, built-in tools, thinking levels, and migration details.

Review Gemini 3.5 Flash-Lite pricing, 1M context, 64K output, multimodal inputs, thinking levels, tools, and high-throughput use cases.

Review Laguna S 2.1 specs: 118B-A8B MoE architecture, 1M context, OpenMDW-1.1 license, quantized weights, serving options, and limitations.

Review Claude Opus 5 API details: model ID, 1M context, 128K output, adaptive thinking, platform availability, pricing, and migration use cases.

Review VisionPsy-Nano-460M: 460M parameters, 8K context, image understanding, Apache 2.0 weights, Flash variant, device use, and limits.

Review Neutrino-1 8B: 8.19B parameters, 40,960 context, 3.88 GB runtime artifact, Apache 2.0 weights, local serving, and compatibility.

Review the retired DeepSeek V4 Flash 0731 model, its open weights, historical API contract, and legacy ID now routing to V4.1 Flash.

Review LG AI Research's K-EXAONE 2.0 750B-A37B open weights, 37B active parameters, 10-language metadata, Apache 2.0 license, and deployment limits.

Review openPangu 2.0 Pro's 505B total and 18B active parameters, 512K context, thinking modes, Ascend deployment path, weights, and custom license.

Review Qwen3.8-Max model ID, September 2 snapshot aliases, 1M context, multimodal inputs, 131K output, pricing, and API usage.

Review Qwen3.8-2.4T-A95B open weights, 95B active parameters, text-only limits, context length, custom license, and serving paths.

Review Qwen3.8-27B open weights, Apache-2.0 license, multimodal inputs, context limits, QwenCloud pricing, and local serving paths.

Review the retired DeepSeek V4 Flash Vision Exp model and its legacy API ID, which now temporarily routes to DeepSeek V4.1 Flash.

Review Qwen3.8-Flash-Next open weights, Qwen4 architecture preview, 125B-A6B MoE design, multimodal inputs, context limits, license, and serving paths.

Claude Fable 5.1 is Anthropic's generally available model for long-horizon coding and research. Check its API ID, 1M context, pricing, and migration changes.

Claude Mythos 5.1 is Anthropic's trusted-access sibling of Fable 5.1 for approved cybersecurity and life-science work. See access, specs, and pricing.

Explore GPT-6 Astra for coding and editable Blender scenes, with plan-specific access, text/image input, API pricing and tool requirements.

Review Gemini 3.8 Flash's stable API ID, launch pricing, 1M context, 65K output, multimodal inputs, thinking levels, tools, and availability.

Review Grok 4.6's API ID, 500K context, $2/$6 pricing, long-prompt rates, image input, reasoning levels, tools, and provider availability.

DeepSeek V4.1 Flash 1017: unchanged deepseek-flash API and prices, dated MIT weights, image input, 1M context, 384K output and official project demos.

Review Jev 1.13: jev-1.13.0 and jev-latest IDs, typed Choice/Score/Noul output, text-only input, context limits, pricing and decision-model caveats.

Review GPT-6 Sol's API ID, 1.05M context, 128K output, reasoning levels, tools, caching, long-prompt pricing, and availability.

Review GPT-6 Luna's API ID, 1.05M context, 128K output, reasoning levels, tools, caching, long-prompt pricing, and access.

Review Claude Opus 5.5's API ID, 1M context, 128K output, adaptive thinking, pricing, cache rates, and migration fit.

Review Grok 4.7's API ID, 500K context, long-prompt pricing, reasoning levels, image input, US endpoint, and Fast boundary.

Check GPT-6.1 Sol API ID, 1.05M context, 128K output, $2/$10 rates, $0.10 cached input, long-context charges, and availability.

Review Claude Sonnet 5.5's API ID, 1M context, 128K output, $2/$10 token rates, adaptive thinking, and Sonnet 5 migration checks.

Mistral Large 4 is a multimodal open-weight model with 1M context, hosted API access and released FP8/NVFP4 checkpoints.

Anthropic's fast model for classification, extraction and subagent work, with prompt-length-based pricing.

Meta's October 12 model for long-running agents, with API access distinct from the subscription-only Full variant.

Google's frontier model with October 13 paid API and Pro/Ultra app rollout; Fairwind cyber access remains separate.

Check Glimpse’s image/video inputs, hosted API ID, token prices and different OpenRouter and AI Gateway limits. Research-preview model from Elorian.

Check Meta's Muse Spark Open weights, license, hardware and OpenRouter API. Compare provider-specific token limits and conditional commercial terms.

Reflection Beam is an open-weight text reasoning model with 501.3B total and 23.4B active parameters. Compare 1M context, Apache 2.0 terms and checkpoints.

MiniMax M3.1 Flash is a hosted reasoning model with 1M context. Compare API token rates, M Plan and MiniMax Code access; weights remain a separate milestone.

Nemotron 3.5 Super VL has 120B total and 12B active parameters with up to 1M context. Compare multimodal inputs, OpenMDW terms and hardware requirements.

Nemotron 3.5 Bolt is a 74B-total, 9B-active multimodal model with up to 256K context. Compare BF16 and native 2-bit 24GB weights and memory requirements.

Step 5 Preview is a 600B-total, 27B-active model with FP8 and BF16 weights. Check its 1M context, conditional StepFun license and large-memory deployment.

Jev 1.14 is TypeSafe’s versioned System One model for typed decisions from text. Check preview versus latest aliases, default limits and long-context access.

Mercury 3 is Inception’s text-diffusion LLM with an advertised 1M context. Compare its native API, standard token rates and temporary launch discounts.

GPT-6.1 Terra accepts text and images with 1.05M context and 128K output; standard API rates are $1 input, $0.05 cached input and $5 output per million tokens.

Kimi K3.1 Pro: 1,048,576 context and always-on thinking. API $2 input/$10 output; cache writes $2 (5m)/$4 (1h), hits $0.20 per million tokens.

Kimi K3.1 Pro Highspeed: 1,048,576 context and always-on thinking. API $4 input/$20 output; cache writes $4 (5m)/$8 (1h), hits $0.20 per million tokens.

Check Claude Fable 5.2 API IDs, $8/$40 token rates, 1M context, 128K output, thinking settings and migration limits before updating your integration.