97 models
GPT-6.2 Astra
openaiOpenAI's most capable GPT-6.2 model for the hardest long-horizon projects, from start to finish. May 2026 knowledge, 1M context
GPT-6.2 Sol
openaiOpenAI's everyday GPT-6.2 model for coding and agents at scale, June 2026 knowledge. 1M context
GPT-6 Astra
openaiOpenAI flagship — GPT-6 generation, tuned for long-horizon agentic work and computer use. 1M context
GPT-6.1 Sol
openaiRefreshed GPT-6 Sol — complex coding and agentic workflows below the Astra flagship, April 2026 knowledge. 1M context
GPT-6 Sol
openaiCost-efficient GPT-6 tier — complex coding and agentic workflows below the Astra flagship. 1M context
GPT-6 Luna
openaiFast, low-cost GPT-6 tier — high-volume chat, classification and lightweight agentic work. 1M context
GPT-5.6 Sol
openaiTop GPT-5.6 tier — deepest reasoning for complex coding, agentic workflows, and long-horizon problems. 1M context
GPT-5.6 Terra
openaiBalanced GPT-5.6 tier — everyday coding, reasoning, and agentic tasks at half the flagship price. 1M context
GPT-5.6 Luna
openaiCost-efficient GPT-5.6 tier for high-volume, latency-sensitive chat and lightweight agentic workflows. 1M context
GPT-5.6 Sol Pro
openaiHighest-capability GPT-5.6 — Sol with pro reasoning mode for the hardest problems and long-running agentic work. 1M context
GPT-5.6 Terra Pro
openaiGPT-5.6 Terra with pro reasoning mode — deeper responses on complex tasks at half the standard Terra rate. 1M context
GPT-5.6 Luna Pro
openaiGPT-5.6 Luna with pro reasoning mode — budget tier with deeper reasoning for high-volume workloads. 1M context
GPT-5.5
openaiFirst fully retrained base since GPT-4.5. 1M context, 128K output, native agent + computer use
GPT-5.5 Pro
openaiPremium GPT-5.5 with maximum compute for the hardest problems
ChatGPT Instant (GPT-5.5)
openaiChatGPT's default model — the rolling `chat-latest` alias, currently GPT-5.5 Instant. Tuned for speed and concision, same price as GPT-5.5
GPT-5.4
openaiGPT-5.4 frontier model with 1M context, native computer use, and thinking mode
GPT-5.4 Pro
openaiPremium GPT-5.4 with maximum compute for the hardest problems
GPT-5.1
openaiFrontier model with 400K context and configurable reasoning effort
GPT-5.2
openaiFrontier model with 400K context and adaptive reasoning
GPT-5.4 Mini
openaiStrongest mini model for coding, computer use, and subagents with GPT-5.4 capabilities
GPT-5 Mini
openaiCost-optimized reasoning and chat
GPT-5.4 Nano
openaiFastest and most affordable GPT-5.4 model for high-throughput tasks
GPT-5.2 Pro
openaiUses more compute for consistently better answers
GPT-5.3 Codex
openaiIndustry-leading agentic coding model. 400K context, reasoning, tool use, and complex execution
GPT-4.1
openaiLatest GPT-4 generation model
GPT-4.1 Mini
openaiFast and affordable GPT-4.1 model
GPT-4o
openaiMultimodal model with vision and audio
GPT-4o Mini
openaiFast and affordable GPT-4o model
o1
openaiAdvanced reasoning model for complex tasks
o3
openaio-series reasoning model with improved performance
o3-mini
openaiEfficient reasoning model for STEM tasks
o4-mini
openaiEfficient o4-generation reasoning model
Claude Haiku 4.5
anthropicFastest and most efficient Claude, near-frontier intelligence
Claude Sonnet 5.5
anthropicNewest Sonnet — fast, capable everyday coding and agent work at Sonnet cost. 1M context, 128k output, adaptive thinking, vision
Claude Sonnet 5
anthropicSonnet 5 — near-Opus coding/agentic quality at Sonnet cost. 1M context, 128k output, adaptive thinking, vision
Claude Sonnet 4.6
anthropicSonnet 4.6 — a balance of intelligence, speed, and cost
Claude Sonnet 4.5
anthropicSonnet 4.5 — strong coding and agentic performance, vision
Claude Opus 4.5
anthropicClaude Opus 4.5 — enhanced reasoning and creativity
Claude Opus 4.7
anthropicPowerful Claude Opus for complex reasoning and agentic coding. 1M context, 128k output, adaptive thinking
Claude Fable 5.2
anthropicAnthropic's most capable model — successor to Fable 5.1 at a lower price, for the most demanding reasoning and long-horizon agentic work. 1M context, 128K output, always-on thinking. Anthropic retains requests to this model for 30 days.
Claude Fable 5.1
anthropicAnthropic's most capable model — successor to Fable 5 in the same tier at the same price, for the most demanding reasoning and long-horizon agentic work. 1M context, 128K output, always-on thinking
Claude Fable 5
anthropicMythos-class tier above Opus, for the most demanding reasoning and long-horizon agentic work. 1M context, 128K output, always-on thinking
Claude Opus 4.8
anthropicMost capable Claude 4-series Opus for complex reasoning and agentic coding. 1M context, 128k output, adaptive thinking
Claude Opus 5
anthropicOpus 5 — step-change over Opus 4.8 for deep reasoning and agentic coding at the same price. 1M context, 128k output, adaptive thinking
Claude Opus 5.5
anthropicNewest Opus — Opus-class deep reasoning and agentic coding at a lower price than Opus 5. 1M context, 128k output, always-on adaptive thinking, vision
Gemini 4 Carbon
googleGoogle's most capable reasoning model, built for complex software engineering, deep math and science, and long autonomous agent work
Gemini 4 Argon
googleGoogle's mid-tier Gemini 4 model: Gemini 4 intelligence with faster responses and fewer tokens per turn than Carbon
Gemini 4 Neon
googleGoogle's newest Gemini 4 model, built for agentic coding, knowledge work and long-context reasoning, with a 1M-token context window
Gemini 4 Hydrogen
googleGoogle's fast, efficient Gemini 4 model for everyday coding and agentic work, with a 1M-token context window
Gemini 3.1 Pro
googleGemini 3.1 Pro with improved thinking, token efficiency, and agentic capabilities. Optimized for software engineering (requires new SDK)
Gemini 3 Flash Preview
googleFrontier-class performance with Pro-level intelligence at Flash speed and pricing. Includes thinking mode (requires new SDK)
Gemini 3.8 Flash
googleNewest-generation Flash with built-in thinking mode — frontier-class quality at Flash speed
Gemini 3.6 Flash
googleGemini 3.6 Flash with built-in thinking mode — frontier-class quality at Flash speed
Gemini 3.5 Flash
googleGemini 3.5 Flash with built-in thinking mode — frontier-class quality at Flash speed
Gemini 2.5 Pro
googleGemini 2.5 Pro — strong reasoning, coding, and mathematics
Gemini 2.5 Flash
googleFast and efficient Gemini model with vision support
Gemini 3.5 Flash Lite
googleLatest Flash Lite — ultra-fast, lightweight Gemini with thinking mode for high-throughput tasks
Gemini 3.1 Flash Lite
googleUltra-fast and lightweight Gemini 3.1 model with thinking mode for high-throughput tasks
Gemini 2.5 Flash Lite
googleMost economical Gemini model - ultra-fast and lightweight (requires new SDK)
DeepSeek V4 Flash Vision
deepseekDeepSeek V4 Flash Vision — 1M context with text and image input. Experimental multimodal tier of the V4 Flash family.
DeepSeek V4 Pro
deepseekDeepSeek V4 flagship — 1.6T MoE / 49B active, 1M context. Strongest open-weight reasoner. Thinking mode default.
DeepSeek V4.1 Flash
deepseekDeepSeek V4.1 Flash in non-thinking mode — 1M context at $0.14 in / $0.28 out. Production-grade reliability and 5MB request bodies.
DeepSeek Chat (V4.1 Flash)
deepseekCompatibility id for DeepSeek V4.1 Flash in non-thinking mode ($0.14 in / $0.28 out) — same model, price and routing as deepseek/deepseek-v4.1-flash. Production-grade reliability and 5MB request bodies.
DeepSeek Reasoner (V4.1 Flash)
deepseekCompatibility id for DeepSeek V4.1 Flash in thinking mode for reasoning tasks ($0.14 in / $0.28 out). Same upstream as deepseek/deepseek-v4.1-flash but with thinking enabled by default.
Kimi K3.1 Pro
moonshotMoonshot's flagship — post-trained successor to Kimi K3 for long-horizon coding and agents. 1M context, image + text input, returns reasoning_content.
GLM-5.3
zaiZ.AI's flagship — 1M-token context with always-on reasoning, strong at long-horizon coding. Verified live on Z.AI.
GLM-5.3 Flash
zaiZ.AI's first natively multimodal GLM-5 — 320B/18B MoE with image input, 1M-token context, and always-on reasoning at value-tier pricing. Verified live on Z.AI.
GLM-5.2
zaiZ.AI GLM-5.2 — 1M-token context, strong open-source long-horizon coding. Verified live on Z.AI.
GLM-5.1
zaiZ.AI's GLM-5.1 — #1 open source on SWE-Bench Pro, 8-hour autonomous execution. 200K context
GLM-5
zaiZ.AI's foundation model with 200K context. Strong reasoning and agentic capabilities
GLM-5 Turbo
zaiOptimized GLM-5 variant with faster inference
Grok 4.3
xaixAI's Grok 4.3 reasoning model. 1M context, vision-capable, tuned for agentic workflows and instruction-following.
Grok Build 0.1
xaixAI's fast agentic coding model, trained for interactive software-engineering workflows. 256K context, text + image input.
Grok 4.7
xaixAI's flagship Grok 4.7 — 500K context, vision, always-on reasoning with selectable effort (low to xhigh). Supports Live Search (+$0.30 per call)
Grok 4.6
xaixAI's Grok 4.6 — 500K context, chain-of-thought reasoning with selectable effort. Supports Live Search (+$0.30 per call)
Grok 4.5
xaixAI's Grok 4.5 — 500K context, vision-capable, chain-of-thought reasoning. Supports Live Search (+$0.30 per call)
MiniMax M2.7
minimaxMiniMax's M2.7 reasoning model with recursive self-improvement. Great value for complex tasks (~60 tps)
MiniMax M3
minimaxMiniMax's M3 flagship — 1M context, strong reasoning + coding.
Qwen3.8 Max
qwenAlibaba's Qwen3.8 Max — current flagship of the Qwen line. 1M context, 131K output.
Qwen3.7 Max
qwenAlibaba's Qwen3.7 Max tier. 1M context, strong reasoning, coding, and agentic tool use. Live-verified 2026-07-20.
Qwen3.7 Plus
qwenAlibaba's balanced Qwen tier — 1M context with reasoning, coding, and agentic tool use at a fraction of the Max price
Qwen3.7 Flash
qwenAlibaba's fastest Qwen tier — 1M context reasoning for high-volume, latency-sensitive workloads
Qwen3.8 Flash
qwenAlibaba's Qwen3.8 Flash — 125B MoE with hybrid attention, 1M context, image input. Outperforms the Qwen3.7 Plus tier at a lower price.
Tencent Hy4 Preview
tencentTencent's Hy4 — 770B mixture-of-experts, Apache 2.0, 1M context. A large open flagship at a fraction of closed-model pricing.
MiMo V2.5
xiaomiXiaomi's MiMo V2.5 — 310B sparse MoE, natively multimodal, 1M context. Accepts text and images.
Xiaomi MiMo-V2.5 Pro
xiaomiXiaomi's MiMo-V2.5 Pro — 1M context reasoning model, priced well below the frontier tier.
Nemotron 3.5 Super 120B (Free)
nvidiaFreeNVIDIA Nemotron 3.5 Super VL: MoE model (12B active params) with thinking mode. Reads images, fast and capable reasoning
Nemotron 3 Nano Omni (Free)
nvidiaFreeNVIDIA's multimodal reasoning Nemotron Nano Omni, free. 31B / 3.2B active MoE. Accepts text, images, video, audio. ChartQA 90.3, DocVQA 95.6, MMMU 70.8 — the only free model in our catalog that takes video and audio as well as images
Nemotron 3.5 Lightning (Free)
nvidiaFreeNVIDIA Nemotron 3.5 Lightning 30B-A3B, free. Thinking-mode reasoning with a 1M-token context.
Nemotron 3 Ultra 550B (Free)
nvidiaFreeNVIDIA Nemotron 3 Ultra 550B-A55B, free. 550B total / 55B active MoE with a 1M-token context — the largest free model in the catalog.
Cohere North Mini Code (Free)
cohereFreeCohere's North Mini Code, free. Compact coding model, 256K context, sub-second responses.
Poolside Laguna XS 2.1 (Free)
nvidiaFreePoolside Laguna XS 2.1 hosted free by NVIDIA. Fast compact coding model (~161 tok/s).
Laguna S 2.1
poolsidePoolside's Laguna S 2.1 — 118B open-weight coding model, 1M context, very cheap per token.
Mistral Large 4
mistralMistral's flagship — multimodal reasoning model for coding and agentic work. 512K context, image input.
Muse Glimmer 30B (Free)
nvidiaFreeMeta's Muse Glimmer 30B — Apache 2.0, 29.6B dense, hosted free by NVIDIA. Meta's open line after Llama.
Nemotron 3.5 Bolt (Free)
nvidiaFreeNVIDIA's small, fast Nemotron 3.5 hosted free by NVIDIA. Answers in about a second, reads images and calls tools; no thinking pass. 131K context
MiniMax M3.1 Flash (Free)
nvidiaFreeMiniMax's M3.1 Flash hosted free by NVIDIA. Reasons before it answers, calls tools, and takes a 1M context, usually in a few seconds.