Undominated.ai

Reasoning effort

Max effort is a pricing dimension. We report what vendors publish. We do not ship a reasoning-tax τ point estimate — that needs first-party probes with median and IQR, n≥15.

82/364 cannot disable thinking
242 reasoning supported
160 supported and optional

Effort ladders below are vendor labels (low / high / xhigh / max). An Elo without its effort level is not a fact — see methodology.

ModelDefaultEfforts
AionLabs: Aion 3.5Checkhighmax, high, low
AionLabs: Aion 3.5 MiniCheckhighmax, high, low
AionLabs: Aion-2.0Check——
AionLabs: Aion-3.0Check——
AionLabs: Aion-3.0-MiniCheck——
Anthropic: Claude Fable 5Checkhighmax, xhigh, high, medium, low
Anthropic: Claude Fable 5 (batch)highmax, xhigh, high, medium, low
Anthropic: Claude Fable 5.1Checkhighmax, xhigh, high, medium, low
Anthropic: Claude Fable 5.1 (batch)highmax, xhigh, high, medium, low
Anthropic: Claude Opus 5.5Checkhighmax, xhigh, high, medium, low
Anthropic: Claude Opus 5.5 (batch)highmax, xhigh, high, medium, low
Anthropic: Claude Sonnet 5.5Checkhighmax, xhigh, high, medium, low
Anthropic: Claude Sonnet 5.5 (batch)highmax, xhigh, high, medium, low
Arcee AI: Trinity Large ThinkingCheck——
DeepSeek: R1Check——
DeepSeek: R1 0528Check——
Google: Gemini 2.5 ProCheck——
Google: Gemini 2.5 Pro (batch)——
Google: Gemini 2.5 Pro Preview 06-05Check——
Google: Gemini 3.1 Pro PreviewCheckmediumhigh, medium, low
Google: Gemini 3.1 Pro Preview (batch)mediumhigh, medium, low
Google: Gemini 3.1 Pro Preview Custom ToolsCheckmediumhigh, medium, low
Google: Gemini 3.5 FlashCheckmediumhigh, medium, low, minimal
Google: Gemini 3.5 Flash (batch)mediumhigh, medium, low, minimal
Google: Gemini 3.5 Flash LiteCheckminimalhigh, medium, low, minimal
Google: Gemini 3.5 Flash Lite (batch)minimalhigh, medium, low, minimal
Google: Gemini 3.6 FlashCheckmediumhigh, medium, low, minimal
Google: Gemini 3.6 Flash (batch)mediumhigh, medium, low, minimal
Google: Gemini 3.7 FlashCheckmediumhigh, medium, low
Google: Gemini 3.7 Flash (batch)mediumhigh, medium, low
Google: Gemini 3.8 FlashCheckmediumhigh, medium, low
Google: Gemini 3.8 Flash (batch)mediumhigh, medium, low
Google: Nano Banana Pro (Gemini 3 Pro Image Preview)Check——
Google: Nano Banana Pro (Gemini 3 Pro Image)Check——
LiquidAI: LFM2.5-2.6B (free)——
Meta: Muse Glimmer 30BCheckmediumxhigh, high, medium, low
Meta: Muse Spark 1.1Checkmediumxhigh, high, medium, low, minimal
Meta: Muse Spark 1.2Checkmediumxhigh, high, medium, low, minimal
Meta: Muse Spark 1.2 ContributorCheckmediumxhigh, high, medium, low, minimal
Meta: Muse Spark 1.3Checkmediummax, xhigh, high, medium, low, minimal
Meta: Muse Spark 1.3 ContributorCheckmediummax, xhigh, high, medium, low, minimal
MiniMax: MiniMax M2Check——
MiniMax: MiniMax M2.1Check——
MiniMax: MiniMax M2.5Check——
MiniMax: MiniMax M2.7Check——
MoonshotAI: Kimi K2 ThinkingCheck——
MoonshotAI: Kimi K2.7 CodeCheck——
OpenAI: GPT-5Checkmediumhigh, medium, low, minimal
OpenAI: GPT-5 (batch)mediumhigh, medium, low, minimal
OpenAI: GPT-5 ImageCheck——
OpenAI: GPT-5 Image MiniCheck——
OpenAI: GPT-5 MiniCheckmediumhigh, medium, low, minimal
OpenAI: GPT-5 Mini (batch)mediumhigh, medium, low, minimal
OpenAI: GPT-5 NanoCheckmediumhigh, medium, low, minimal
OpenAI: GPT-5 Nano (batch)mediumhigh, medium, low, minimal
OpenAI: GPT-5 ProCheckhighhigh
OpenAI: GPT-5 Pro (batch)highhigh
OpenAI: GPT-5.1-CodexCheckmediumhigh, medium, low
OpenAI: GPT-5.1-Codex-MaxCheckmediumxhigh, high, medium, low
OpenAI: GPT-5.2 ProCheckmediumxhigh, high, medium
OpenAI: GPT-5.2 Pro (batch)mediumxhigh, high, medium
OpenAI: GPT-5.2-CodexCheckmediumxhigh, high, medium, low
OpenAI: GPT-5.4 ProCheckmediumxhigh, high, medium
OpenAI: GPT-5.4 Pro (batch)mediumxhigh, high, medium
OpenAI: GPT-5.5 ProCheckmediumxhigh, high, medium
OpenAI: GPT-5.5 Pro (batch)mediumxhigh, high, medium
OpenAI: GPT-6 AstraCheckmediummax, xhigh, high, medium, low
OpenAI: GPT-6 Astra (batch)mediummax, xhigh, high, medium, low
OpenAI: GPT-6 Astra ProCheckmediummax, xhigh, high, medium, low
OpenAI: GPT-6 Astra Pro (batch)mediummax, xhigh, high, medium, low
OpenAI: GPT-6.1 SolCheckmediummax, xhigh, high, medium, low
OpenAI: GPT-6.1 Sol ProCheckmediummax, xhigh, high, medium, low
OpenAI: gpt-oss-120bCheckmediumhigh, medium, low
OpenAI: gpt-oss-120b (batch)mediumhigh, medium, low
OpenAI: gpt-oss-20bCheckmediumhigh, medium, low
OpenAI: gpt-oss-20b (batch)mediumhigh, medium, low
OpenAI: gpt-oss-safeguard-20bCheck——
OpenAI: o3 Mini HighCheckhighhigh
OpenAI: o4 Mini HighCheckhighhigh
Perplexity: Sonar Pro SearchCheck——
Qwen: Qwen3 235B A22B Thinking 2507Check——
Qwen: Qwen3 30B A3B Thinking 2507Check——
Qwen: Qwen3 Next 80B A3B ThinkingCheck——
Qwen: Qwen3 VL 235B A22B ThinkingCheck——
Qwen: Qwen3 VL 30B A3B ThinkingCheck——
Qwen: Qwen3 VL 8B ThinkingCheck——
Qwen: Qwen3.8 2.4T A95BCheckxhighxhigh, medium, low
Qwen: Qwen3.8 Max (0902)Checkxhighxhigh, high, medium, low, minimal
Qwen: Qwen3.8 Max PrimeCheckxhighxhigh, high, medium, low, minimal
Reka Flash 3Check——
Sakana: Fugu MaxCheckhighmax, xhigh, high
Sakana: Fugu UltraCheckxhighmax, xhigh, high
Sakana: Fugu Ultra v2Checkxhighmax, xhigh, high
SpaceXAI: Grok 4.20 Multi-AgentCheckmediumxhigh, high, medium, low
SpaceXAI: Grok 4.5Checkhighhigh, medium, low
SpaceXAI: Grok 4.6Checkhighxhigh, high, medium, low
SpaceXAI: Grok 4.7Checkhighxhigh, high, medium, low
SpaceXAI: Grok Build 0.1Check——
StepFun: Step 3.5 FlashCheck——
StepFun: Step 3.7 FlashCheckmediumhigh, medium, low
Z.ai: GLM 5.3Checkmaxmax, high, low
Z.ai: GLM 5.3 (batch)maxmax, high, low
Z.ai: GLM 5.3 FlashCheckmaxmax, high, low
Z.ai: GLM 5.3 Flash (batch)maxmax, high, low
Z.ai: GLM 5.3 FlashXCheckmaxmax, high, low
Z.ai: GLM 5.3 PrimeCheckmaxmax, high, low

All 106 mandatory-reasoning entries are shown, including delivery variants. Check is available for standard-delivery models.

Input hashes & method version

Read citation text and BibTeX
82 of 364 publishable models (23%) declare mandatory reasoning: 106 of 447 catalogue entries, delivery variants included. 242 models declare reasoning support; 160 support optional reasoning. Vendor declarations are not a measured reasoning-cost estimate. As of 2026-10-06. Lens: Vendor reasoning declarations in the accepted OpenRouter-derived catalogue. https://undominated.ai/effort/
Method: https://undominated.ai/methodology/
Publication: sha256:ecef83f767cd4d2c21766f6a5b4dffa36c06351f23657542d473bc1a030c0932; calculation SHA-256: 5ae4d0bee9b70298dd437b05df87c32acb407040b50bf88240677f295f25955d; permitted inputs: https://undominated.ai/data/publications/manifests/ecef83f767cd4d2c21766f6a5b4dffa36c06351f23657542d473bc1a030c0932.json

@misc{undominated-effort-2026-10-06,
  author = {{Undominated.ai}},
  title  = {effort evidence — Undominated.ai},
  year   = {2026},
  url    = {https://undominated.ai/effort/},
  note   = {as of 2026-10-06; Vendor reasoning declarations in the accepted OpenRouter-derived catalogue},
  annote = {Publication: sha256:ecef83f767cd4d2c21766f6a5b4dffa36c06351f23657542d473bc1a030c0932; calculation SHA-256: 5ae4d0bee9b70298dd437b05df87c32acb407040b50bf88240677f295f25955d; permitted inputs: https://undominated.ai/data/publications/manifests/ecef83f767cd4d2c21766f6a5b4dffa36c06351f23657542d473bc1a030c0932.json}
}
Evidence & Ask
Markdown for LLMs