Claude Sonnet 5 vs GPT-5.6 Sol
Claude Sonnet 5 is no longer on the leaderboard. GPT-5.6 Sol was superseded by GPT-6.1 Sol. The grades below are the last ones measured, so read them as history rather than as this week's board. See the current roster.
GPT-5.6 Sol leads overall, C- to D. Claude Sonnet 5 is about 1.5x cheaper per task. By suite, GPT-5.6 Sol takes real-world issues (C- to D+) and spec planning (C+ to D). Take Claude Sonnet 5 anyway when you want the side that is cheaper per task.Measured
Each agent graded on the same private suites, running headless through its own tools. Advantages are oriented so a win is a win, whichever way the metric runs.
Verdict: GPT-5.6 Sol vs Claude Sonnet 5
GPT-5.6 Sol leads overall, C- to D. By suite: GPT-5.6 Sol takes real-world issues, C- to D+ (60% vs 53% capability); GPT-5.6 Sol takes spec planning, C+ to D (68% vs 44% capability); GPT-5.6 Sol takes vibe coding, D+ to F (51% vs 31% capability). Claude Sonnet 5 used 2% of Claude Max 5x's weekly limit per pass, $0.024 per task at the listed price. GPT-5.6 Sol used 3% of ChatGPT Pro 5x's weekly limit per pass, $0.036 per task at the listed price. Per task, Claude Sonnet 5 is about 1.5x cheaper ($0.024 vs $0.036). GPT-5.6 Sol is faster on real-world issues: a median task takes 3m 13s against 3m 35s. Both finished every task on it. Reasons to pick Claude Sonnet 5 anyway: it is cheaper per task ($0.024 vs $0.036).
Changed since the previous graded week: Claude Sonnet 5: vibe coding D+ to F; GPT-5.6 Sol: vibe coding C- to D+.
| Dimension | Claude Sonnet 5 | GPT-5.6 Sol | Advantage |
|---|---|---|---|
| Capability | 53% | 60% | +7% (GPT-5.6 Sol leads) |
| Median task time | 3m 35s | 3m 13s | 22s faster (GPT-5.6 Sol leads) |
| Approx. task cost | $0.015 / task | $0.031 / task | $0.015 / task lower (Claude Sonnet 5 leads) |
Claude Sonnet 5: 12.0% of Claude Max 5x (12.0% / 5h, 1.0% / 1w) per run. Its weekly limit holds about 13.8 of the week's 33.6 5h windows. GPT-5.6 Sol: 2.0% of ChatGPT Pro 5x (2.0% / 1w) per run. Each figure is a share of that agent's own plan, so there is no delta to draw.
| Dimension | Claude Sonnet 5 | GPT-5.6 Sol | Advantage |
|---|---|---|---|
| Capability | 44% | 68% | +25% (GPT-5.6 Sol leads) |
| Median task time | 5m 12s | 6m 0s | 48s faster (Claude Sonnet 5 leads) |
| Approx. task cost | $0.057 / task | $0.057 / task | Even |
Claude Sonnet 5: 5.0% of Claude Max 5x (5.0% / 5h, 1.0% / 1w) per run. Its weekly limit holds about 13.8 of the week's 33.6 5h windows. GPT-5.6 Sol: 1.0% of ChatGPT Pro 5x (1.0% / 1w) per run. Each figure is a share of that agent's own plan, so there is no delta to draw.
| Dimension | Claude Sonnet 5 | GPT-5.6 Sol | Advantage |
|---|---|---|---|
| Capability | 31% | 51% | +20% (GPT-5.6 Sol leads) |
| Median task time | 4m 31s | 3m 56s | 35s faster (GPT-5.6 Sol leads) |
| Approx. task cost | $0.007 / task | $0.092 / task | $0.085 / task lower (Claude Sonnet 5 leads) |
Claude Sonnet 5: 1.0% of Claude Max 5x (1.0% / 5h) per good solve. Its weekly limit holds about 13.8 of the week's 33.6 5h windows. GPT-5.6 Sol: 0.4% of ChatGPT Pro 5x (0.4% / 1w) per good solve. Each figure is a share of that agent's own plan, so there is no delta to draw.
Grades and deltas are comparable only within a suite version. See methodology.
Frequently asked questions
Which is better for coding, Claude Sonnet 5 or GPT-5.6 Sol?
GPT-5.6 Sol clears higher: tier 4 of 5 versus tier 3 for Claude Sonnet 5 on real-world@v3, our private suite of real-world coding problems. Latest measurement 2026-09-28.
Which is cheaper per task, Claude Sonnet 5 or GPT-5.6 Sol?
Claude Sonnet 5 used 2% of Claude Max 5x's weekly limit per pass, $0.024 per task at the listed price. GPT-5.6 Sol used 3% of ChatGPT Pro 5x's weekly limit per pass, $0.036 per task at the listed price. Per task, Claude Sonnet 5 is about 1.5x cheaper ($0.024 vs $0.036).
What do Claude Sonnet 5 and GPT-5.6 Sol cost a month?
Claude Sonnet 5 runs on Claude Max 5x at $100/month (public price as of 2026-07-20). GPT-5.6 Sol runs on ChatGPT Pro 5x at $100/month (public price as of 2026-09-05).