The Best AI Models of February 2026
Opus 4.6, Gemini 3.1 Pro, GPT-5.3-Codex, Grok 4.20, and more!
February was a blockbuster month for model releases, with Opus 4.6, Gemini 3.1 Pro, GPT-5.3-Codex, Grok 4.20 and more.
February 2026 Model Releases
Opus 4.6 and Gemini 3.1 Pro, released February 6th and February 19th, respectively, each pushed model frontiers to new heights, earning them the right to share the title of the new all-around state-of-the-art models and re-defining what it means to be Tier 1 as of February 2026.
Beyond these two, it is worth giving special mention to GPT-5.3-Codex and Grok 4.20.
GPT-5.3-Codex was released on February 6th on the same days as Opus 4.6 and it is a clear step above GPT-5.2. That said, we didn’t see OpenAI release a general purpose GPT-5.3 for some reason and we didn’t see GPT-5.3-Codex get as much benchmark coverage as GPT-5.2. So while GPT-5.3-Codex seems able to be competitive with Opus 4.6 and Gemini 3.1 Pro, it is a bit harder to judge this model vs previous GPT releases.
Grok 4.20, meanwhile, was a massive step up for xAI from its previous frontier model Grok 4. Its capabilities launched xAI back into the running and this rate of improvement shows that xAI is still behind the other 3 big US AI labs but it is gaining ground and not losing it.
As far as the Chinese labs go, we saw the following releases:
February 12th: GLM 5
February 12th: MiniMax M2.5
February 16th: Qwen3.5-397B
Updated Model Rankings
Tier 1
Gemini-3.1-Pro (New)
Opus-4.6 (New)
GPT-5.3-Codex (New)
Tier 2
Grok-4.20 (New)
Kimi K2.5
GLM-5 (New)
DeepSeek-V3.2
MiniMax M2.5 (New)
Qwen3.5-397B (New)
Looking Forward
In a follow-up post, we’ll dig into the benchmarks of the various AI models released this month and assess them on IQ, EQ, coding, math, science, and several other critical dimensions.
In the meantime you can dig into the data and view the AI model graphs on aiiq.org.
Stay tuned for more.