As OpenCode gains adoption as a flexible, provider-agnostic alternative to Codex CLI and its competitors, we benchmarked the two head-to-head on OpenAI’s fastest coding model (GPT 5.1 Codex Mini) and their flagship, most accurate coding model (GPT 5.2 Codex).

What we measured:

• Sigmascore — the overall measure of an agent’s real-world coding performance
• Accuracy — how often outputs meet quality thresholds
• Consistency — how often outputs remain useful even when not fully completing a task
• Speed — how quickly tasks are completed

Each score is assigned a tier based on how close they are with respect to the margin of error. The top-scoring group of agents are in Tier 1, the next-best group are in Tier 2, and so on.

Codex CLI outperforms OpenCode across both GPT-5.1 Codex Mini and GPT-5.2 Codex:

• On GPT-5.1 Codex Mini, OpenCode ranks 12th overall (Tier 6) on the Sigmabench leaderboard, while Codex CLI ranks 11th (Tier 5), due to a measurable speed disadvantage.
• On GPT-5.2 Codex, both agents move up the Sigmabench leaderboard, with OpenCode placing 8th overall (Tier 4) and Codex CLI 4th overall (Tier 3).
• Across both models, Accuracy and Consistency are statistically tied, but Codex CLI’s faster median runtimes are sufficient to lift it one full Sigmascore tier ahead of OpenCode.

Comparisons

OpenCode vs Codex CLI (GPT 5.1 Codex Mini)

In this comparison, OpenCode performs slightly worse than Codex CLI overall. It is marginally worse in Accuracy (-1.1 points), marginally better in Consistency (+0.3 points), and quite a bit worse in Speed (-2.8 points).

From a user’s perspective, OpenCode will run GPT-5.1 Codex Mini just as well as Codex CLI in terms of output quality, but it will do so about 22% slower (616s vs 505s median runtimes). This difference in speed drives the OpenCode Sigmascore down a full tier compared to Codex CLI (Tier 8 vs Tier 7).

OpenCode vs Codex CLI (GPT 5.2 Codex)

On OpenAI’s flagship coding model, GPT 5.2 Codex, our benchmark tells a similar story as with GPT 5.1 Codex Mini. OpenCode is statistically tied with Codex CLI on Accuracy, but lags behind a full tier on Consistency and is once again a bit slower.

This time, however, the gap in Speed is smaller: the median runtime was 474s for OpenCode and 420s for Codex CLI, a difference of about 13%. Once again, this brings OpenCode’s Sigmascore down a full tier compared to Codex CLI (Tier 7 vs Tier 6).

Median Runtime Comparison

The following chart recaps the median runtime differences between OpenCode and Codex CLI across both models. On both GPT 5.1 Codex Mini and GPT 5.2 Codex, OpenCode is slower than Codex CLI, with slightly wider 95% CIs for OpenCode, indicating a slightly higher variance in runtime.

OpenCode vs Codex CLI - Key Insights

• OpenCode and Codex CLI are statistically tied on Accuracy across both GPT 5.1 Codex Mini and GPT 5.2 Codex.

• OpenCode is statistically tied with Codex CLI on Consistency for GPT 5.1 Codex Mini but trails behind a full tier on GPT 5.2 Codex.

• OpenCode is slower than Codex CLI across both models, with a median runtime difference of about 22% for GPT 5.1 Codex Mini and about 13% for GPT 5.2 Codex.

• OpenCode’s Sigmascore is down a full tier compared to Codex CLI across both models due to the speed difference.