Best models for agents
Models that support the tools parameter for function calling. 245 of 338 catalog models match as of July 2026.
- 245
- Models
- $0.010
- Input from /1M
- 10M
- Max context
- Jun 30, 2026
- Newest release
245 models in the OpenKey catalog fit the agents facet as of now, with Claude Sonnet 5 (Anthropic, released 2026-06-30) the newest entry — a 1M-context model at $2/$10 per M input/output. Pricing spans from free tiers (Nemotron 3 Ultra free, Laguna M.1 free, Laguna XS.2 free) up to $30/$180 per M on GPT-5.5 Pro and Claude Opus 4.7 (Fast). Most of the field clusters in the $0.1-$5 input range with 256K-1M context, which is the range that actually matters for agent loops: you're paying per tool call and per retry, not per single completion. A handful of models — GLM 5.2, Claude Opus 4.8, Claude Fable 5 — carry design_arena benchmark ranks worth checking before you commit a workflow to one backbone.
Top picks
Claude Sonnet 5
Newest model in the set (2026-06-30), 1M context window, and priced at $2/$10 per M in/out — mid-range for a frontier-class model, which matters when an agent loop burns tokens on tool-call retries and long scratchpads.
GLM 5.2
Ranks 1 on design_arena 3d (elo 1376), the highest elo in this list, at $0.93/$3 per M with a 1,048,576-token context window — strong reasoning signal for a fraction of Opus-tier pricing.
Nemotron 3 Ultra (free)
Same 1M context and design_arena 3d rank 40 (elo 1213) as its paid twin, but $0/$0 per M — useful for prototyping an agent's tool-calling logic before you swap in a paid model for production runs.
All 245 models
Full catalog →Claude Sonnet 5Anthropic1Mctx 1M · inout
Fugu UltraSakana AI1Mctx 1M · inout
Nano Banana Pro (Gemini 3 Pro Image)Google66Kctx 66K · inout
North Mini Code (free)FreeCohere256KFreeFreectx 256K · inFreeoutFree
GLM 5.2Z.ai1.0Mctx 1.0M · inout
Kimi K2.7 CodeMoonshot AI262Kctx 262K · inout
Claude Fable 5Anthropic1Mctx 1M · inout
Nemotron 3 UltraNVIDIA1Mctx 1M · inout
Nemotron 3 Ultra (free)FreeNVIDIA1MFreeFreectx 1M · inFreeoutFree
Qwen3.7 PlusQwen1Mctx 1M · inout
MiniMax M3MiniMax1.0Mctx 1.0M · inout
Step 3.7 FlashStepFun256Kctx 256K · inout
Claude Opus 4.8Anthropic1Mctx 1M · inout
Claude Opus 4.8 (Fast)Anthropic1Mctx 1M · inout
Qwen3.7 MaxQwen1Mctx 1M · inout
Grok Build 0.1xAI256Kctx 256K · inout
Gemini 3.5 FlashGoogle1.0Mctx 1.0M · inout
Claude Opus 4.7 (Fast)Anthropic1Mctx 1M · inout
Ring-2.6-1TInclusionAI262Kctx 262K · inout
Gemini 3.1 Flash LiteGoogle1.0Mctx 1.0M · inout
GPT Chat LatestOpenAI400Kctx 400K · inout
Granite 4.1 8BIBM131Kctx 131K · inout
Mistral Medium 3.5Mistral AI262Kctx 262K · inout
Grok 4.3xAI1Mctx 1M · inout
Nemotron 3 Nano Omni (free)FreeNVIDIA256KFreeFreectx 256K · inFreeoutFree
Laguna M.1poolside262Kctx 262K · inout
Laguna M.1 (free)Freepoolside262KFreeFreectx 262K · inFreeoutFree
Laguna XS.2poolside262Kctx 262K · inout
Laguna XS.2 (free)Freepoolside262KFreeFreectx 262K · inFreeoutFree
Qwen3.5 Plus 2026-04-20Qwen1Mctx 1M · inout
Qwen3.6 27BQwen262Kctx 262K · inout
Qwen3.6 35B A3BQwen262Kctx 262K · inout
Qwen3.6 FlashQwen1Mctx 1M · inout
Qwen3.6 Max PreviewQwen262Kctx 262K · inout
DeepSeek V4 FlashDeepSeek1.0Mctx 1.0M · inout
DeepSeek V4 ProDeepSeek1.0Mctx 1.0M · inout
GPT-5.5OpenAI1.1Mctx 1.1M · inout
GPT-5.5 ProOpenAI1.1Mctx 1.1M · inout
Ling-2.6-1TInclusionAI262Kctx 262K · inout
Hy3 previewTencent262Kctx 262K · inout
MiMo-V2.5Xiaomi1.0Mctx 1.0M · inout
MiMo-V2.5-ProXiaomi1.0Mctx 1.0M · inout
Ling-2.6-flashInclusionAI262Kctx 262K · inout
Kimi K2.6Moonshot AI262Kctx 262K · inout
Claude Opus 4.7Anthropic1Mctx 1M · inout
GLM 5.1Z.ai203Kctx 203K · inout
Gemma 4 26B A4B Google262Kctx 262K · inout
Gemma 4 26B A4B (free)FreeGoogle262KFreeFreectx 262K · inFreeoutFree
Gemma 4 31BGoogle262Kctx 262K · inout
Gemma 4 31B (free)FreeGoogle262KFreeFreectx 262K · inFreeoutFree
Qwen3.6 PlusQwen1Mctx 1M · inout
Trinity Large ThinkingArcee AI262Kctx 262K · inout
GLM 5V TurboZ.ai203Kctx 203K · inout
Grok 4.20xAI2Mctx 2M · inout
KAT-Coder-Pro V2Kwaipilot256Kctx 256K · inout
Reka EdgeReka16Kctx 16K · inout
MiniMax M2.7MiniMax205Kctx 205K · inout
GPT-5.4 MiniOpenAI400Kctx 400K · inout
GPT-5.4 NanoOpenAI400Kctx 400K · inout
Mistral Small 4Mistral AI262Kctx 262K · inout
GLM 5 TurboZ.ai262Kctx 262K · inout
Nemotron 3 SuperNVIDIA1Mctx 1M · inout
Nemotron 3 Super (free)FreeNVIDIA1MFreeFreectx 1M · inFreeoutFree
Seed-2.0-LiteByteDance Seed262Kctx 262K · inout
Qwen3.5-9BQwen262Kctx 262K · inout
GPT-5.4OpenAI1.1Mctx 1.1M · inout
GPT-5.4 ProOpenAI1.1Mctx 1.1M · inout
Mercury 2Inception128Kctx 128K · inout
Gemini 3.1 Flash Lite PreviewGoogle1.0Mctx 1.0M · inout
GPT-5.3 ChatOpenAI128Kctx 128K · inout
Seed-2.0-MiniByteDance Seed262Kctx 262K · inout
Gemini 3.1 Pro Preview Custom ToolsGoogle1.0Mctx 1.0M · inout
Qwen3.5-122B-A10BQwen262Kctx 262K · inout
Qwen3.5-27BQwen262Kctx 262K · inout
Qwen3.5-35B-A3BQwen262Kctx 262K · inout
Qwen3.5-FlashQwen1Mctx 1M · inout
GPT-5.3-CodexOpenAI400Kctx 400K · inout
Gemini 3.1 Pro PreviewGoogle1.0Mctx 1.0M · inout
Claude Sonnet 4.6Anthropic1Mctx 1M · inout
Qwen3.5 397B A17BQwen256Kctx 256K · inout
Qwen3.5 Plus 2026-02-15Qwen1Mctx 1M · inout
MiniMax M2.5MiniMax205Kctx 205K · inout
GLM 5Z.ai203Kctx 203K · inout
Qwen3 Max ThinkingQwen262Kctx 262K · inout
Claude Opus 4.6Anthropic1Mctx 1M · inout
Qwen3 Coder NextQwen262Kctx 262K · inout
Free Models RouterFreeOpenRouter200KFreeFreectx 200K · inFreeoutFree
Step 3.5 FlashStepFun262Kctx 262K · inout
Kimi K2.5Moonshot AI262Kctx 262K · inout
Solar Pro 3Upstage128Kctx 128K · inout
LFM2.5-1.2B-Thinking (free)FreeLiquid AI33KFreeFreectx 33K · inFreeoutFree
GPT AudioOpenAI128Kctx 128K · inout
GPT Audio MiniOpenAI128Kctx 128K · inout
GLM 4.7 FlashZ.ai203Kctx 203K · inout
GPT-5.2-CodexOpenAI400Kctx 400K · inout
Seed 1.6ByteDance Seed262Kctx 262K · inout
Seed 1.6 FlashByteDance Seed262Kctx 262K · inout
MiniMax M2.1MiniMax205Kctx 205K · inout
GLM 4.7Z.ai203Kctx 203K · inout
Gemini 3 Flash PreviewGoogle1.0Mctx 1.0M · inout
Nemotron 3 Nano 30B A3BNVIDIA262Kctx 262K · inout
Nemotron 3 Nano 30B A3B (free)FreeNVIDIA256KFreeFreectx 256K · inFreeoutFree
GPT-5.2OpenAI400Kctx 400K · inout
GPT-5.2 ChatOpenAI128Kctx 128K · inout
GPT-5.2 ProOpenAI400Kctx 400K · inout
Devstral 2 2512Mistral AI262Kctx 262K · inout
Relace SearchRelace256Kctx 256K · inout
GLM 4.6VZ.ai131Kctx 131K · inout
GPT-5.1-Codex-MaxOpenAI400Kctx 400K · inout
Nova 2 LiteAmazon1Mctx 1M · inout
Ministral 3 14B 2512Mistral AI262Kctx 262K · inout
Ministral 3 3B 2512Mistral AI131Kctx 131K · inout
Ministral 3 8B 2512Mistral AI262Kctx 262K · inout
Trinity MiniArcee AI131Kctx 131K · inout
DeepSeek V3.2DeepSeek131Kctx 131K · inout
Mistral Large 3 2512Mistral AI262Kctx 262K · inout
Claude Opus 4.5Anthropic200Kctx 200K · inout
GPT-5.1OpenAI400Kctx 400K · inout
GPT-5.1 ChatOpenAI128Kctx 128K · inout
GPT-5.1-CodexOpenAI400Kctx 400K · inout
GPT-5.1-Codex-MiniOpenAI400Kctx 400K · inout
Kimi K2 ThinkingMoonshot AI262Kctx 262K · inout
Nova Premier 1.0Amazon1Mctx 1M · inout
Voxtral Small 24B 2507Mistral AI32Kctx 32K · inout
gpt-oss-safeguard-20bOpenAI131Kctx 131K · inout
Nemotron Nano 12B 2 VL (free)FreeNVIDIA128KFreeFreectx 128K · inFreeoutFree
MiniMax M2MiniMax205Kctx 205K · inout
Qwen3 VL 32B InstructQwen262Kctx 262K · inout
Claude Haiku 4.5Anthropic200Kctx 200K · inout
Qwen3 VL 8B InstructQwen256Kctx 256K · inout
Qwen3 VL 8B ThinkingQwen256Kctx 256K · inout
Llama 3.3 Nemotron Super 49B V1.5NVIDIA131Kctx 131K · inout
o3 Deep ResearchOpenAI200Kctx 200K · inout
o4 Mini Deep ResearchOpenAI200Kctx 200K · inout
GPT-5 ProOpenAI400Kctx 400K · inout
Qwen3 VL 30B A3B InstructQwen262Kctx 262K · inout
Qwen3 VL 30B A3B ThinkingQwen131Kctx 131K · inout
GLM 4.6Z.ai203Kctx 203K · inout
Claude Sonnet 4.5Anthropic1Mctx 1M · inout
DeepSeek V3.2 ExpDeepSeek164Kctx 164K · inout
Gemini 2.5 Flash Lite Preview 09-2025Google1.0Mctx 1.0M · inout
GPT-5 CodexOpenAI400Kctx 400K · inout
Qwen3 Coder PlusQwen1Mctx 1M · inout
Qwen3 MaxQwen262Kctx 262K · inout
Qwen3 VL 235B A22B InstructQwen262Kctx 262K · inout
Qwen3 VL 235B A22B ThinkingQwen131Kctx 131K · inout
DeepSeek V3.1 TerminusDeepSeek164Kctx 164K · inout
Qwen3 Coder FlashQwen1Mctx 1M · inout
Qwen3 Next 80B A3B InstructQwen262Kctx 262K · inout
Qwen3 Next 80B A3B Instruct (free)FreeQwen262KFreeFreectx 262K · inFreeoutFree
Qwen3 Next 80B A3B ThinkingQwen262Kctx 262K · inout
Qwen Plus 0728Qwen1Mctx 1M · inout
Qwen Plus 0728 (thinking)Qwen1Mctx 1M · inout
Nemotron Nano 9B V2 (free)FreeNVIDIA128KFreeFreectx 128K · inFreeoutFree
Kimi K2 0905Moonshot AI262Kctx 262K · inout
Qwen3 30B A3B Thinking 2507Qwen131Kctx 131K · inout
DeepSeek V3.1DeepSeek164Kctx 164K · inout
Mistral Medium 3.1Mistral AI131Kctx 131K · inout
GLM 4.5VZ.ai66Kctx 66K · inout
Jamba Large 1.7AI21 Labs256Kctx 256K · inout
GPT-5OpenAI400Kctx 400K · inout
GPT-5 MiniOpenAI400Kctx 400K · inout
GPT-5 NanoOpenAI400Kctx 400K · inout
Claude Opus 4.1Anthropic200Kctx 200K · inout
gpt-oss-120bOpenAI131Kctx 131K · inout
gpt-oss-120b (free)FreeOpenAI131KFreeFreectx 131K · inFreeoutFree
gpt-oss-20bOpenAI131Kctx 131K · inout
gpt-oss-20b (free)FreeOpenAI131KFreeFreectx 131K · inFreeoutFree
Codestral 2508Mistral AI256Kctx 256K · inout
Qwen3 Coder 30B A3B InstructQwen160Kctx 160K · inout
Qwen3 30B A3B Instruct 2507Qwen131Kctx 131K · inout
Qwen3 235B A22B Thinking 2507Qwen262Kctx 262K · inout
GLM 4.5Z.ai131Kctx 131K · inout
GLM 4.5 AirZ.ai131Kctx 131K · inout
Qwen3 Coder 480B A35BQwen1.0Mctx 1.0M · inout
Qwen3 Coder 480B A35B (free)FreeQwen1.0MFreeFreectx 1.0M · inFreeoutFree
Gemini 2.5 Flash LiteGoogle1.0Mctx 1.0M · inout
Qwen3 235B A22B Instruct 2507Qwen262Kctx 262K · inout
Kimi K2 0711Moonshot AI131Kctx 131K · inout
Mistral Small 3.2 24BMistral AI128Kctx 128K · inout
Gemini 2.5 FlashGoogle1.0Mctx 1.0M · inout
Gemini 2.5 ProGoogle1.0Mctx 1.0M · inout
MiniMax M1MiniMax1Mctx 1M · inout
o3 ProOpenAI200Kctx 200K · inout
Gemini 2.5 Pro Preview 06-05Google1.0Mctx 1.0M · inout
R1 0528DeepSeek164Kctx 164K · inout
Claude Opus 4Anthropic200Kctx 200K · inout
Claude Sonnet 4Anthropic1Mctx 1M · inout
Gemini 2.5 Pro Preview 05-06Google1.0Mctx 1.0M · inout
Mistral Medium 3Mistral AI131Kctx 131K · inout
Virtuoso LargeArcee AI131Kctx 131K · inout
Qwen3 14BQwen132Kctx 132K · inout
Qwen3 235B A22BQwen131Kctx 131K · inout
Qwen3 30B A3BQwen131Kctx 131K · inout
Qwen3 32BQwen131Kctx 131K · inout
Qwen3 8BQwen131Kctx 131K · inout
o3OpenAI200Kctx 200K · inout
o4 MiniOpenAI200Kctx 200K · inout
o4 Mini HighOpenAI200Kctx 200K · inout
GPT-4.1OpenAI1.0Mctx 1.0M · inout
GPT-4.1 MiniOpenAI1.0Mctx 1.0M · inout
GPT-4.1 NanoOpenAI1.0Mctx 1.0M · inout
Llama 4 MaverickMeta AI1.0Mctx 1.0M · inout
Llama 4 ScoutMeta AI10Mctx 10M · inout
DeepSeek V3 0324DeepSeek164Kctx 164K · inout
Gemma 3 12BGoogle131Kctx 131K · inout
Gemma 3 27BGoogle131Kctx 131K · inout
SabaMistral AI33Kctx 33K · inout
o3 Mini HighOpenAI200Kctx 200K · inout
Qwen-PlusQwen1Mctx 1M · inout
o3 MiniOpenAI200Kctx 200K · inout
R1DeepSeek164Kctx 164K · inout
DeepSeek V3DeepSeek131Kctx 131K · inout
o1OpenAI200Kctx 200K · inout
Llama 3.3 70B InstructMeta AI131Kctx 131K · inout
Llama 3.3 70B Instruct (free)FreeMeta AI131KFreeFreectx 131K · inFreeoutFree
Nova Lite 1.0Amazon300Kctx 300K · inout
Nova Micro 1.0Amazon128Kctx 128K · inout
Nova Pro 1.0Amazon300Kctx 300K · inout
GPT-4o (2024-11-20)OpenAI128Kctx 128K · inout
Mistral Large 2407Mistral AI131Kctx 131K · inout
UnslopNemo 12BTheDrummer33Kctx 33K · inout
Qwen2.5 7B InstructQwen131Kctx 131K · inout
Qwen2.5 72B InstructQwen131Kctx 131K · inout
Command R (08-2024)Cohere128Kctx 128K · inout
Command R+ (08-2024)Cohere128Kctx 128K · inout
Llama 3.1 Euryale 70B v2.2Sao10K131Kctx 131K · inout
GPT-4o (2024-08-06)OpenAI128Kctx 128K · inout
Llama 3.1 70B InstructMeta AI131Kctx 131K · inout
Llama 3.1 8B InstructMeta AI131Kctx 131K · inout
Mistral NemoMistral AI131Kctx 131K · inout
GPT-4o-miniOpenAI128Kctx 128K · inout
GPT-4o-mini (2024-07-18)OpenAI128Kctx 128K · inout
GPT-4oOpenAI128Kctx 128K · inout
GPT-4o (2024-05-13)OpenAI128Kctx 128K · inout
Mixtral 8x22B InstructMistral AI66Kctx 66K · inout
GPT-4 TurboOpenAI128Kctx 128K · inout
Claude 3 HaikuAnthropic200Kctx 200K · inout
Mistral LargeMistral AI128Kctx 128K · inout
GPT-3.5 Turbo (older v0613)OpenAI4Kctx 4K · inout
GPT-4 Turbo PreviewOpenAI128Kctx 128K · inout
Auto RouterOpenRouter2MVariableVariablectx 2M · inVariableoutVariable
GPT-3.5 Turbo 16kOpenAI16Kctx 16K · inout
GPT-3.5 TurboOpenAI16Kctx 16K · inout
GPT-4OpenAI8Kctx 8K · inout
Prices per 1M tokens, flat 3% fee included. Hover or tap a price for the math.
How to choose
Agent workloads punish you twice for the wrong pick: on cost (multi-step loops multiply token spend) and on context (a 262K window like Kimi K2.7 Code's fills up fast once you add tool schemas and history). GLM 5.2 and DeepSeek V4 Pro give you 1M+ context at under $1 input while ranking on design_arena — good defaults for long-running agents. If you need top-tier reasoning and can absorb the cost, Claude Opus 4.8 ranks 1 on design_arena androidnative (elo 1343) at $5/$25 per M. For throwaway prototyping, the free tier (Nemotron 3 Ultra free, Laguna M.1 free) costs nothing but ships the same benchmark numbers as their paid counterparts.
Questions
- What's the cheapest model here that still has agent-relevant benchmarks?
- Ring-2.6-1T and Ling-2.6-1T both sit at $0.075/$0.625 per M with 262,144-token context, though neither lists a best_benchmark score. If you want a priced, benchmarked option, DeepSeek V4 Flash is $0.09/$0.18 per M and ranks on design_arena svg (elo 1216).
- Is there a free model with a real context window for agent testing?
- Yes — Nemotron 3 Ultra (free) and North Mini Code (free) both run at $0/$0. Nemotron 3 Ultra (free) gives you the full 1,000,000-token context plus a design_arena 3d rank 40 (elo 1213), matching its paid version exactly.
- How does OpenKey pricing work on these models?
- You pay the provider's list price plus a flat 3% fee. For Claude Sonnet 5 at $2 input / $10 output per M tokens, that's $2 × 1.03 = $2.06 input and $10 × 1.03 = $10.30 output on OpenKey — same model, one key, no separate provider accounts.