GPT-5.3 Codex vs GPT-5.4 vs Kimi K3

It is close. Our weighted score puts them within a point (Kimi K3 65/100, GPT-5.4 65/100, GPT-5.3 Codex 64/100), so choose by what matters most for your work: Kimi K3 for raw capability and GPT-5.3 Codex on price. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.

GPT-5.3 Codex is cheaper at $1.75 input / $14.00 output per million tokens (official OpenAI API price). GPT-5.4 costs $2.50 input / $15.00 output per million tokens (official OpenAI API price); Kimi K3 costs $3.00 input / $15.00 output per million tokens (official Moonshot AI API price). At a typical mix of three input tokens to one output token, that is $4.81 per million tokens for GPT-5.3 Codex versus $5.63 for GPT-5.4 (1.2× as much) and $6.00 for Kimi K3 (1.2× as much).

Kimi K3 scores higher on the Capabilities Index (ECI): Kimi K3 157.6 (#13 of 148), GPT-5.4 156.9 (#16 of 148) and GPT-5.3 Codex 156.8 (#18 of 148). The confidence ranges of the top two overlap (154.9–160.4 vs 154.9–159.1), so treat the gap as small.

There are no published SWE-bench Verified results for Kimi K3 yet, so there is no like-for-like coding score. On overall capability, Kimi K3 leads, which tends to carry over to coding, but test on your own codebase. All three support tool calling for agent workflows.

GPT-5.4 and Kimi K3 have the largest context windows (1,050,000 and 1,048,576 tokens), against 400,000 for GPT-5.3 Codex. Maximum output per response: GPT-5.3 Codex up to 128,000, GPT-5.4 up to 128,000, Kimi K3 up to 131,072 tokens.

GPT-5.3 Codex accepts text, images and PDFs; GPT-5.4 accepts text, images and PDFs; Kimi K3 accepts text, images and video. They handle the same number of input types.

Kimi K3 publishes its weights and can be self-hosted; GPT-5.3 Codex and GPT-5.4 is proprietary.

Kimi K3 is the newest, released Jul 16, 2026. GPT-5.4 came out Mar 5, 2026; GPT-5.3 Codex came out Feb 5, 2026. Knowledge cutoff: GPT-5.3 Codex Aug 31, 2025, GPT-5.4 Aug 31, 2025.

Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.