Best coding model right now

Measured inside the coding tools themselves, never scored on a benchmark.

45 models ranked · 12 boards · captured 3 August 2026 · against 2 August 2026

Overall Free 10 Under $1 / M 10 $1 to $5 / M 10 Above $5 / M 10 By tool 7

Best coding model overall — every eligible model, ranked on the volume routed through the tracked coding tools. A cheap model can lead on volume alone, and often does.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	Z.ai: GLM 5.2z-ai/glm-5.2	$1.83	1.0M	Claude Code Codex CLI+4	2.88T	Held
2	StepFun: Step 3.7 Flashstepfun/step-3.7-flash	$0.44	262K	Cline Kilo Code	2.70T	Held
3	Tencent: Hy3tencent/hy3	$0.23	262K	Claude Code Codex CLI+4	1.80T	Held
4	DeepSeek: DeepSeek V4 Flashdeepseek/deepseek-v4-flash	$0.17	1.0M	Claude Code Codex CLI+4	652B	Held
5	Xiaomi: MiMo-V2.5xiaomi/mimo-v2.5	$0.17	1.1M	Claude Code Codex CLI+4	338B	Held
6	NVIDIA: Nemotron 3 Ultranvidia/nemotron-3-ultra-550b-a55b	$1.35	512K	Claude Code Codex CLI+4	928B	Held
7	NVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:free	Free	1M	Claude Code Codex CLI+4	928B	Held
8	Poolside: Laguna M.1poolside/laguna-m.1	not published	262K	Claude Code Aider+3	1.12T	Held
9	MiniMax: MiniMax M3minimax/minimax-m3	$0.53	1.0M	Claude Code Cline+2	696B	Held
10	Anthropic: Claude Opus 4.8anthropic/claude-opus-4.8	$10.00	1M	Claude Code Codex CLI+3	862B	Held

By what you are willing to spend

4 bands

Measured — counted from real repositories, nightlyAsserted — read from the vendor docs, dated

Best in Free

10 models

Models published at no per-token cost. Free to route through, which is why volume here is not the same signal as volume in a paid band.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	NVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:free	Free	1M	Claude Code Codex CLI+4	928B	Held
2	Ling-3.0-flash (free)inclusionai/ling-3.0-flash:free	Free	262K	Aider Kilo Code+1	433B	Held
3	Poolside: Laguna S 2.1 (free)poolside/laguna-s-2.1:free	Free	262K	Cline Kilo Code	454B	Held
4	NVIDIA: Nemotron 3 Super (free)nvidia/nemotron-3-super-120b-a12b:free	Free	262K	Kilo Code Roo Code	188B	Held
5	Poolside: Laguna XS 2.1 (free)poolside/laguna-xs-2.1:free	Free	262K	Cline Kilo Code	191B	Held
6	Google: Gemma 4 31B (free)google/gemma-4-31b-it:free	Free	262K	Aider	157M	Held
7	Cohere: North Mini Code (free)cohere/north-mini-code:free	Free	256K	Roo Code	1B	Held
8	NVIDIA: Nemotron 3 Nano Omni (free)nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free	Free	256K	Kilo Code	32B	New
9	NVIDIA: Nemotron 3 Nano 30B A3B (free)nvidia/nemotron-3-nano-30b-a3b:free	Free	256K	Roo Code	1B	1
10	Google: Gemma 4 26B A4B (free)google/gemma-4-26b-a4b-it:free	Free	262K	not attributed	372B	2

Best in Under $1 / M

10 models

Under a dollar per million tokens blended. Where most agent traffic ends up once someone is paying for their own keys.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	StepFun: Step 3.7 Flashstepfun/step-3.7-flash	$0.44	262K	Cline Kilo Code	2.70T	1
2	Tencent: Hy3tencent/hy3	$0.23	262K	Claude Code Codex CLI+4	1.80T	1
3	DeepSeek: DeepSeek V4 Flashdeepseek/deepseek-v4-flash	$0.17	1.0M	Claude Code Codex CLI+4	652B	1
4	Xiaomi: MiMo-V2.5xiaomi/mimo-v2.5	$0.17	1.1M	Claude Code Codex CLI+4	338B	1
5	MiniMax: MiniMax M3minimax/minimax-m3	$0.53	1.0M	Claude Code Cline+2	696B	1
6	DeepSeek: DeepSeek V4 Prodeepseek/deepseek-v4-pro	$0.54	1.0M	Claude Code Codex CLI+4	316B	1
7	Poolside: Laguna S 2.1poolside/laguna-s-2.1	$0.11	1.0M	Cline Kilo Code	454B	2
8	DeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731	$0.11	1.0M	Claude Code Codex CLI+1	261B	New
9	Xiaomi: MiMo-V2.5-Proxiaomi/mimo-v2.5-pro	$0.54	1.1M	Cline Kilo Code+1	343B	1
10	OpenAI: GPT-5.6 Lunaopenai/gpt-5.6-luna	$0.23	1.1M	Claude Code Codex CLI	141B	New

Best in $1 to $5 / M

10 models

One to five dollars per million tokens blended.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	Z.ai: GLM 5.2z-ai/glm-5.2	$1.83	1.0M	Claude Code Codex CLI+4	2.88T	New
2	NVIDIA: Nemotron 3 Ultranvidia/nemotron-3-ultra-550b-a55b	$1.35	512K	Claude Code Codex CLI+4	928B	1
3	Anthropic: Claude Sonnet 5anthropic/claude-sonnet-5	$4.00	1M	Claude Code Aider+3	677B	1
4	Anthropic: Claude Haiku 4.5anthropic/claude-haiku-4.5	$2.00	200K	Claude Code Aider	86B	1
5	xAI: Grok 4.5x-ai/grok-4.5	$3.00	500K	Codex CLI Roo Code	10B	1
6	MoonshotAI: Kimi K2.7 Codemoonshotai/kimi-k2.7-code	$1.42	262K	Cline Roo Code	11B	New
7	OpenAI: GPT-5.6 Terraopenai/gpt-5.6-terra	$2.25	1.1M	Codex CLI	28B	1
8	Google: Gemini 3.5 Flashgoogle/gemini-3.5-flash	$3.38	1.0M	Claude Code	52B	1
9	Google: Gemini 3 Flash Previewgoogle/gemini-3-flash-preview	$1.13	1.0M	not attributed	983B	1
10	OpenAI: GPT-5.4 Miniopenai/gpt-5.4-mini	$1.69	400K	Codex CLI	8B	1

Best in Above $5 / M

10 models

Above five dollars per million tokens blended. Small volumes here still mean real spend.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	Anthropic: Claude Opus 4.8anthropic/claude-opus-4.8	$10.00	1M	Claude Code Codex CLI+3	862B	Held
2	MoonshotAI: Kimi K3moonshotai/kimi-k3	$6.00	1.0M	Claude Code Codex CLI+4	257B	Held
3	Anthropic: Claude Sonnet 4.6anthropic/claude-sonnet-4.6	$6.00	1M	Claude Code Aider+3	276B	Held
4	Claude Opus 5anthropic/claude-opus-5	$10.00	1M	Claude Code Cline	290B	1
5	Anthropic: Claude Fable 5anthropic/claude-fable-5	$20.00	1M	Claude Code Cline	388B	1
6	Anthropic: Claude Opus 4.6anthropic/claude-opus-4.6	$10.00	1M	Claude Code Aider+1	108B	Held
7	OpenAI: GPT-5.6 Solopenai/gpt-5.6-sol	$11.25	1.1M	Codex CLI	242B	Held
8	Anthropic: Claude Opus 4.7anthropic/claude-opus-4.7	$10.00	1M	Claude Code	111B	Held
9	OpenAI: GPT-5.6 Sol Proopenai/gpt-5.6-sol-pro	$11.25	1.1M	Codex CLI	91B	Held
10	OpenAI: GPT-5.5openai/gpt-5.5	$11.25	1.1M	Codex CLI	54B	Held

Best model inside each tool

7 tools

Measured per tool, so these are not slices of the overall board — a model can lead here and be nowhere near the top of it.

Best model in Aider

10 models

Ordered by each model's share of the tokens Aider sends through OpenRouter — with the bundled default first where there is one. No benchmark is involved.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	DeepSeek: DeepSeek V4 Flashdeepseek/deepseek-v4-flash	$0.17	1.0M	Aider	765M	Held
2	NVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:free	Free	1M	Aider	606M	Held
3	DeepSeek: DeepSeek V4 Prodeepseek/deepseek-v4-pro	$0.54	1.0M	Aider	536M	Held
4	Anthropic: Claude Sonnet 5anthropic/claude-sonnet-5	$4.00	1M	Aider	467M	1
5	Z.ai: GLM 5.2z-ai/glm-5.2	$1.83	1.0M	Aider	462M	1
6	Tencent: Hy3tencent/hy3	$0.23	262K	Aider	423M	Held
7	DeepSeek: DeepSeek V3deepseek/deepseek-chat	$0.45	164K	Aider	380M	Held
8	Xiaomi: MiMo-V2.5xiaomi/mimo-v2.5	$0.17	1.1M	Aider	335M	Held
9	MoonshotAI: Kimi K3moonshotai/kimi-k3	$6.00	1.0M	Aider	262M	1
10	Anthropic: Claude Sonnet 4anthropic/claude-sonnet-4	$6.00	1M	Aider	260M	1

Best model in Claude Code

10 models

Ordered by each model's share of the tokens Claude Code sends through OpenRouter — with the bundled default first where there is one. No benchmark is involved.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	Anthropic: Claude Opus 4.8anthropic/claude-opus-4.8	$10.00	1M	Claude Code	747B	Held
2	Z.ai: GLM 5.2z-ai/glm-5.2	$1.83	1.0M	Claude Code	2.66T	Held
3	Anthropic: Claude Sonnet 5anthropic/claude-sonnet-5	$4.00	1M	Claude Code	477B	Held
4	Anthropic: Claude Fable 5anthropic/claude-fable-5	$20.00	1M	Claude Code	363B	Held
5	Tencent: Hy3tencent/hy3	$0.23	262K	Claude Code	325B	Held
6	DeepSeek: DeepSeek V4 Flashdeepseek/deepseek-v4-flash	$0.17	1.0M	Claude Code	310B	Held
7	NVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:free	Free	1M	Claude Code	268B	Held
8	Claude Opus 5anthropic/claude-opus-5	$10.00	1M	Claude Code	263B	Held
9	Anthropic: Claude Sonnet 4.6anthropic/claude-sonnet-4.6	$6.00	1M	Claude Code	208B	Held
10	MoonshotAI: Kimi K3moonshotai/kimi-k3	$6.00	1.0M	Claude Code	176B	Held

Rank 1 is the model this tool ships with, so it is what you get without configuring anything. Every rank below it is openrouter bring-your-own-key traffic for this tool — a different population, not a runner-up.

Best model in Cline

10 models

Ordered by each model's share of the tokens Cline sends through OpenRouter — with the bundled default first where there is one. No benchmark is involved.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	StepFun: Step 3.7 Flashstepfun/step-3.7-flash	$0.44	262K	Cline	759B	Held
2	Poolside: Laguna M.1poolside/laguna-m.1	not published	262K	Cline	600B	Held
3	DeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731	$0.11	1.0M	Cline	206B	1
4	DeepSeek: DeepSeek V4 Flashdeepseek/deepseek-v4-flash	$0.17	1.0M	Cline	159B	1
5	Tencent: Hy3tencent/hy3	$0.23	262K	Cline	139B	Held
6	Poolside: Laguna S 2.1 (free)poolside/laguna-s-2.1:free	Free	262K	Cline	133B	Held
7	NVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:free	Free	1M	Cline	123B	Held
8	Xiaomi: MiMo-V2.5-Proxiaomi/mimo-v2.5-pro	$0.54	1.1M	Cline	116B	Held
9	Anthropic: Claude Sonnet 5anthropic/claude-sonnet-5	$4.00	1M	Cline	101B	Held
10	Xiaomi: MiMo-V2.5xiaomi/mimo-v2.5	$0.17	1.1M	Cline	91B	Held

Best model in Codex CLI

10 models

Ordered by each model's share of the tokens Codex CLI sends through OpenRouter — with the bundled default first where there is one. No benchmark is involved.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	OpenAI: GPT-5.5openai/gpt-5.5	$11.25	1.1M	Codex CLI	54B	Held
2	OpenAI: GPT-5.6 Solopenai/gpt-5.6-sol	$11.25	1.1M	Codex CLI	242B	Held
3	OpenAI: GPT-5.6 Sol Proopenai/gpt-5.6-sol-pro	$11.25	1.1M	Codex CLI	91B	Held
4	OpenAI: GPT-5.6 Lunaopenai/gpt-5.6-luna	$0.23	1.1M	Codex CLI	90B	Held
5	OpenAI: GPT-5.6 Luna Proopenai/gpt-5.6-luna-pro	$0.23	1.1M	Codex CLI	76B	Held
6	OpenAI: GPT-5.6 Terraopenai/gpt-5.6-terra	$2.25	1.1M	Codex CLI	28B	Held
7	OpenAI: GPT-5.4openai/gpt-5.4	$5.63	1.1M	Codex CLI	15B	Held
8	Tencent: Hy3tencent/hy3	$0.23	262K	Codex CLI	13B	Held
9	DeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731	$0.11	1.0M	Codex CLI	10B	New
10	Z.ai: GLM 5.2z-ai/glm-5.2	$1.83	1.0M	Codex CLI	10B	1

Rank 1 is the model this tool ships with, so it is what you get without configuring anything. Every rank below it is openrouter bring-your-own-key traffic for this tool — a different population, not a runner-up.

Best model in Kilo Code

10 models

Ordered by each model's share of the tokens Kilo Code sends through OpenRouter — with the bundled default first where there is one. No benchmark is involved.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	StepFun: Step 3.7 Flashstepfun/step-3.7-flash	$0.44	262K	Kilo Code	1.94T	Held
2	Tencent: Hy3tencent/hy3	$0.23	262K	Kilo Code	1.31T	Held
3	MiniMax: MiniMax M3minimax/minimax-m3	$0.53	1.0M	Kilo Code	609B	Held
4	NVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:free	Free	1M	Kilo Code	518B	Held
5	Poolside: Laguna M.1poolside/laguna-m.1	not published	262K	Kilo Code	447B	Held
6	Ling-3.0-flash (free)inclusionai/ling-3.0-flash:free	Free	262K	Kilo Code	431B	Held
7	Poolside: Laguna S 2.1 (free)poolside/laguna-s-2.1:free	Free	262K	Kilo Code	321B	Held
8	Xiaomi: MiMo-V2.5-Proxiaomi/mimo-v2.5-pro	$0.54	1.1M	Kilo Code	217B	Held
9	NVIDIA: Nemotron 3 Super (free)nvidia/nemotron-3-super-120b-a12b:free	Free	262K	Kilo Code	187B	Held
10	Poolside: Laguna XS 2.1 (free)poolside/laguna-xs-2.1:free	Free	262K	Kilo Code	175B	Held

Best model in Roo Code

10 models

Ordered by each model's share of the tokens Roo Code sends through OpenRouter — with the bundled default first where there is one. No benchmark is involved.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	DeepSeek: DeepSeek V4 Flashdeepseek/deepseek-v4-flash	$0.17	1.0M	Roo Code	25B	Held
2	Tencent: Hy3tencent/hy3	$0.23	262K	Roo Code	17B	Held
3	DeepSeek: DeepSeek V4 Prodeepseek/deepseek-v4-pro	$0.54	1.0M	Roo Code	17B	Held
4	Z.ai: GLM 5.2z-ai/glm-5.2	$1.83	1.0M	Roo Code	13B	Held
5	NVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:free	Free	1M	Roo Code	12B	Held
6	Anthropic: Claude Sonnet 4.6anthropic/claude-sonnet-4.6	$6.00	1M	Roo Code	10B	1
7	Xiaomi: MiMo-V2.5-Proxiaomi/mimo-v2.5-pro	$0.54	1.1M	Roo Code	10B	1
8	Xiaomi: MiMo-V2.5xiaomi/mimo-v2.5	$0.17	1.1M	Roo Code	9B	Held
9	Anthropic: Claude Opus 4.6anthropic/claude-opus-4.6	$10.00	1M	Roo Code	6B	Held
10	MiniMax: MiniMax M3minimax/minimax-m3	$0.53	1.0M	Roo Code	6B	Held

Best model in Codex Cloud

1 model

Ordered by each model's share of the tokens Codex Cloud sends through OpenRouter — with the bundled default first where there is one. No benchmark is involved.
Rank	Model	Per million	Context	Routed through	Volume	Movement
1	OpenAI: GPT-5.5openai/gpt-5.5	$11.25	1.1M	Codex Cloud	—	Held

What this board measures, and what it does not

It measures: Adoption — how much a model is actually run inside the tracked coding tools.
It does not measure: Capability. No benchmark is involved. A cheap model can lead on volume alone.
Price basis: Blended USD per million tokens, weighting prompt to completion 3:1, because coding traffic is prompt-heavy.
Movement: Each rank is compared against the 2 August 2026 capture of the same board. A missed night means no comparison rather than a comparison against a stale capture.

Read the full method What the tools changed