vs
13:5
overall capability wins (0 ties across 18 scored). GPT-5.6 Sol leads overall.
- Reasoning5:1
- Coding3:2
- Arena0:1
- Agents2:0
- Tool use0:1
- Math1:0
- Composite indices2:0
- Pricing0:3
- Speed2:0
- Specs2:0
Verdict
GPT-5.6 Sol leads coding (widest gap: +6.4 on CursorBench); GPT-5.6 Sol offers larger context (1.1M vs 1.0M); GPT-5.6 Sol allows ~2.0x more max output (128K vs 64K).
- GPT-5.6 Sol leads coding (widest gap: +6.4 on CursorBench)
- GPT-5.6 Sol offers larger context (1.1M vs 1.0M)
- GPT-5.6 Sol allows ~2.0x more max output (128K vs 64K)
- Kimi K3 ships open weights (Kimi K3 License)
- Kimi K3 is ~1.3x cheaper on a blended token basis than GPT-5.6 Sol
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- GPT-5.6 Sol
- $9
- Kimi K3
- $6.75
Kimi K3 is about 1.3x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
OpenAI
Moonshot
Family
GPT-5.x
Kimi
License
Proprietary
Kimi K3 License
Open weights
No
Yes
Release
Jul 9, 2026
Jul 16, 2026
Knowledge cutoff
2026-02-16
-
API / provider
OpenAI
Moonshot
Modalities
text, image → text
text, image, video → text
Specs
Context window
1.1M
1.0M
Max output
128K
64K
Parameters
—
2.8T (104B act.)
Pricing
Input $/1M
$4
$3
Output $/1M
$20
$15
Blended $/1M (3∶1)
$8
$6
Speed
tok/s
96
60
TTFT (s)
0.45
0.6
Reasoning
MMLU-Pro
—
—
GPQA Diamond
94.1
93.5
Humanity's Last Exam
49.5
46.9
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
59.4
47.6
AA-LCR v1.1
84.0
88.7
CritPt
32.3
23.4
MMMU-Pro
83.4
80.5
IFBench
72.7
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
64.6
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
88.0
85.0
Aider Polyglot
—
—
Terminal-Bench 3
34.6
—
BigCodeBench
—
—
SciCode
57.1
59.5
CursorBench
67.2
60.8
SWE-Rebench
62.3
—
NL2Repo-Bench
—
—
DeepSWE
72.7
69.0
WebDev Arena
1,617
1,674
Terminal-Bench 4.0
37.3
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,484
1,485
| Metric | GPT-5.6 Sol | Kimi K3 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | OpenAI | Moonshot | — |
| Family | GPT-5.x | Kimi | — |
| License | Proprietary | Kimi K3 License | — |
| Open weights | No | Yes | — |
| Release | Jul 9, 2026 | Jul 16, 2026 | — |
| Knowledge cutoff | 2026-02-16 | - | — |
| API / provider | OpenAI | Moonshot | — |
| Modalities | text, image → text | text, image, video → text | — |
| Specs | |||
| Context window | 1.1M | 1.0M | A +1K |
| Max output | 128K | 64K | A +64K |
| Parameters | — | 2.8T (104B act.) | — |
| Pricing | |||
| Input $/1M | $4 | $3 | B +$1 |
| Output $/1M | $20 | $15 | B +$5 |
| Blended $/1M (3∶1) | $8 | $6 | B +$2 |
| Speed | |||
| tok/s | 96 | 60 | A +36 |
| TTFT (s) | 0.45 | 0.6 | A +0.15 |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 94.1 | 93.5 | A +0.6 pts |
| Humanity's Last Exam | 49.5 | 46.9 | A +2.6 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 59.4 | 47.6 | A +11.8 pts |
| AA-LCR v1.1 | 84.0 | 88.7 | B +4.7 pts |
| CritPt | 32.3 | 23.4 | A +8.9 pts |
| MMMU-Pro | 83.4 | 80.5 | A +2.9 pts |
| IFBench | 72.7 | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | 64.6 | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | 88.0 | 85.0 | A +3.0 pts |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | 34.6 | — | — |
| BigCodeBench | — | — | — |
| SciCode | 57.1 | 59.5 | B +2.4 pts |
| CursorBench | 67.2 | 60.8 | A +6.4 pts |
| SWE-Rebench | 62.3 | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | 72.7 | 69.0 | A +3.7 pts |
| WebDev Arena | 1,617 | 1,674 | B +57 Elo |
| Terminal-Bench 4.0 | 37.3 | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,484 | 1,485 | B +1 Elo |
GPT-5.6 Sol
- Context
- 1.1M / 128K out
- Parameters
- —
- Price
- $4 / $20
- Speed
- 96 tok/s · 0.45s TTFT
- Modalities
- text, image → text
- License
- Proprietary
OpenAI's July 2026 flagship reasoning tier, tuned for hard STEM and agentic work; ChatGPT's top API SKU in the 5.6 family. Promotional Standard rates are $4/$20 (through at least Nov 21, 2026). Fast mode (2× price) delivers up to 2.5× Standard speed, and the Cerebras-powered Ultrafast preview reaches up to 750 tok/s.
Kimi K3
- Context
- 1.0M / 64K out
- Parameters
- 2.8T (104B act.)
- Price
- $3 / $15
- Speed
- 60 tok/s · 0.6s TTFT
- Modalities
- text, image, video → text
- License
- Kimi K3 License
Largest open-weight model (2.8T MoE / 104B active) — official card reports strong reasoning and agentic coding results.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.