Which model should you pick?
Filter by legal jurisdiction, budget, context and vision. The ranking uses published prices, each model official positioning, context size and the speed our probes actually measure.
Ranking
| # | Version | Provider | Data jurisdiction | Price $/M (in / out) | Measured speed (first token) | Context | Our probe | Score |
|---|---|---|---|---|---|---|---|---|
| 1 | Claude Haiku 5.5 | Anthropic | US | 0,1 / 0,5 USD | — | 1 M | Not probed (no key) | 83,4 |
| 2 | GPT 6 Luna | OpenAI | US | 0,1 / 0,5 USD | — | 1,05 M | Not probed (no key) | 82,5 |
| 3 | Gemini 3.1 Flash-Lite | US | 0,25 / 1,5 USD | 128 ms | 1,05 M | Operational | 51,8 | |
| 4 | Gemini 3.5 Flash-Lite | US | 0,3 / 2,5 USD | 126 ms | 1,05 M | Operational | 46,4 | |
| 5 | Gemini 3.8 Flash | US | 0,75 / 3,75 USD | 239 ms | 1,05 M | Operational | 44,6 | |
| 6 | Grok 4.3 | xAI | US | 1,25 / 2,5 USD | — | 1 M | Not probed (no key) | 43,2 |
| 7 | Grok 4.20 multi-agent | xAI | US | 1,25 / 2,5 USD | — | 1 M | Not probed (no key) | 42,8 |
| 8 | Grok 4.20 reasoning | xAI | US | 1,25 / 2,5 USD | — | 1 M | Not probed (no key) | 42,8 |
| 9 | Grok 4.20 non-reasoning | xAI | US | 1,25 / 2,5 USD | — | 1 M | Not probed (no key) | 41,2 |
| 10 | Gemini 3.6 Flash | US | 0,75 / 3,75 USD | — | 1,05 M | Not probed (no key) | 40,4 | |
| 11 | Grok build-0.1 | xAI | US | 1 / 2 USD | — | 256 k | Not probed (no key) | 40,1 |
| 12 | Gemini 4 Argon | US | 2 / 10 USD | — | 1,05 M | Not probed (no key) | 39,9 |
Why this ranking
1. Claude Haiku 5.5 — Anthropic (US) · Volume, automatisations, prototypage
Weighting : Cost 100 · Speed 45 · Capability 55 · Context 99 · Our probe 60
2. GPT 6 Luna — OpenAI (US) · Fort volume, coût minimal
Weighting : Cost 100 · Speed 45 · Capability 50 · Context 100 · Our probe 60
3. Gemini 3.1 Flash-Lite — Google (US) · Tâches légères à haute fréquence, extraction de données
Weighting : Cost 34 · Speed 98 · Capability 42 · Context 100 · Our probe 100
Score 0-100: cost (input price 30%, output 70%), capability (official positioning in the range), context (log scale), speed (measured first token; neutral 45/100 when not probed), state (100 operational, 55 degraded, 0 down, 60 not measured).