Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models760Deals30d251Policy30d61, 11 pendingSources1,057
Pro Mode — raw scores, one benchmark under the lensthird-party eval harness
The field · 27 measured
4.3379.67

27 models measured, most on the third-party eval harness. Muse Wonder 1.0 tops the board at 79.67.

27 measured·10 new this week·Lifecycle Updated Oct 16, 2026Score basis
#◆Vendor$/M in$/M out
Leader — ignites whiteField — fill cools with scoreBars show 4.33–79.67 · ◆ solid = first-party or better · outlined = arm's-length
01Muse Wonder 1.0NEWMeta79.67———
02Muse Spark 1.3NEWMeta66.89$1.3/M$4.3/M$0.041
03GPT-6 AstraNEWOpenAI65.32$10.0/M$50.0/M$0.459
04GPT-6.1 SolNEWOpenAI62.33$2.0/M$10.0/M$0.096
05Claude Fable 5.1Anthropic61.5$10.0/M$50.0/M$0.488
06Claude Opus 5.5NEWAnthropic59$4.0/M$20.0/M$0.203
07GLM-5.3NEWZ.ai58.33$1.4/M$4.4/M$0.050
08Claude Opus 5Anthropic57$5.0/M$25.0/M$0.263
09Claude Fable 5Anthropic56.33$10.0/M$50.0/M$0.533
10DeepSeek-V4-FlashNEWDeepSeek47.81$0.01/M$1.3/M$0.013
11GLM-5.2Z.ai43.67$1.4/M$4.4/M$0.066
12Kimi K3NEWMoonshot42.67$0.30/M$12.3/M$0.148
13Claude Opus 4.7Anthropic41.67$5.0/M$25.0/M$0.360
14Gemini 3.8 FlashGoogle41.47$0.75/M$3.8/M$0.054
15GPT-5.5OpenAI39.67$5.0/M$30.0/M$0.441
16DeepSeek-V4-ProNEWDeepSeek38.46$0.43/M$0.87/M$0.017
17Claude Opus 4.6Anthropic38.33$5.0/M$25.0/M$0.391
18MiniMax M3NEWMiniMax38.26$0.28/M$1.1/M$0.018
19Gemini 3.1 ProGoogle35.33$2.0/M$12.0/M$0.198
20Claude Opus 4.8Anthropic35.33$5.0/M$25.0/M$0.425
21GPT-5.6 SolOpenAI32.33$4.0/M$20.0/M$0.371
22Gemini 3.5 FlashGoogle27.67$1.5/M$9.0/M$0.190
23Grok 4.20SpaceXAI20$1.3/M$2.5/M$0.094
24Kimi K2.6Moonshot18.67$0.95/M$4.0/M$0.133
25GPT-5.4OpenAI9.67$2.5/M$15.0/M$0.905
26MiniMax M2.5MiniMax6.33$0.27/M$0.95/M$0.096
27GPT-5.3 CodexOpenAI4.33$1.8/M$14.0/M$1.819
◆ solid = independently verified · aggregator attested · vendor attributed — outlined = cross-reference · source-attributed. Dimmed ◆ = confidence inferred. “+N alts” = alternate disclosures in the drawer with why-demoted. Prices per million tokens; $/point = input price ÷ score. Every row opens its model dossier.

Verification: 27 models scored · 27 independently verified · 0 with source disagreement. How these tiers are assigned