GPT-5 Mini
VS
Kimi K2.6
GPT-5 Mini vs Kimi K2.6: which wins at real work?
22 task areas · same graded test runs · rank comparison only, so 0–100 and Elo collections never mix raw scores.
Kimi K2.6 wins 20 of 22 task areas we tested; GPT-5 Mini takes 2. GPT-5 Mini costs 1.8× less per token ($2.25 vs $4.07 per 1M).
GPT-5 Mini costs 1.8× less per token ($2.25 vs $4.07 per 1M).
Task by task
| Task area | GPT-5 Mini | Kimi K2.6 | Winner |
|---|---|---|---|
| Customer Support |
#104
/ 113
Usable
|
#4
/ 113
Strong
|
Kimi K2.6 |
| Executive Assistant |
#108
/ 112
Usable
|
#8
/ 112
Strong
|
Kimi K2.6 |
| Sales |
#102
/ 110
Needs editing
|
#15
/ 110
Strong
|
Kimi K2.6 |
| AI Strategy |
#100
/ 126
Usable
|
#19
/ 126
Strong
|
Kimi K2.6 |
| Summarization & Meeting Notes |
#81
/ 110
Excellent
|
#3
/ 110
Excellent
|
Kimi K2.6 |
| RAG, Safety & Grounding |
#78
/ 113
Excellent
|
#18
/ 113
Excellent
|
Kimi K2.6 |
| Creative & Comedy | #70 / 110 | #11 / 110 | Kimi K2.6 |
| Product & Project Management |
#1
/ 110
Excellent
|
#53
/ 110
Strong
|
GPT-5 Mini |
| Investor & Pitch |
#50
/ 66
Usable
|
#1
/ 66
Strong
|
Kimi K2.6 |
| Content & Brand |
#69
/ 124
Strong
|
#22
/ 124
Strong
|
Kimi K2.6 |
| Chef / Home Cooking |
#45
/ 126
Usable
|
#7
/ 126
Strong
|
Kimi K2.6 |
| Presentations & Decks |
#39
/ 110
Excellent
|
#2
/ 110
Excellent
|
Kimi K2.6 |
| Training & Education |
#81
/ 110
Strong
|
#45
/ 110
Excellent
|
Kimi K2.6 |
| Coding |
#37
/ 115
Strong
|
#5
/ 115
Excellent
|
Kimi K2.6 |
| Translation & Localization |
#52
/ 110
Excellent
|
#22
/ 110
Excellent
|
Kimi K2.6 |
| Landing Pages |
#50
/ 72
Usable
|
#21
/ 72
Strong
|
Kimi K2.6 |
| Knowledge & Docs |
#41
/ 110
Strong
|
#22
/ 110
Strong
|
Kimi K2.6 |
| Legal & HR |
#10
/ 110
Excellent
|
#22
/ 110
Excellent
|
GPT-5 Mini |
| Data & Analytics |
#27
/ 110
Excellent
|
#23
/ 110
Excellent
|
Kimi K2.6 |
| Frontend & Landing Pages |
#12
/ 109
Needs editing
|
#8
/ 109
Usable
|
Kimi K2.6 |
| Structured Output |
#17
/ 113
Excellent
|
#14
/ 113
Excellent
|
Kimi K2.6 |
| Research & Competitive Analysis |
#21
/ 110
Excellent
|
#19
/ 110
Excellent
|
Kimi K2.6 |
Rank = position among every model config we tested in that task area (lower is better). Sorted by biggest gap first.
Same task, both models — judged
Both models answered the same test case; an independent judge graded each. Quotes are the judge's actual rationale.
Customer Support
Login lockout (Lumen) (Resolution)
“The model violates the hard gate by inventing policy (that failed attempts extend the lock) and making an unauthorized promise (an immediate phone callback). These details are not present in the provided ground truth.”
“The model perfectly follows the provided policy, acknowledges the urgency with empathy, and provides clear, actionable steps for the customer to regain access. It correctly offers both the manual unlock via identity verification and the 15-minute auto-clear option without inventing any policy or fabricating account facts.”
Summarization & Meeting Notes
Northwind leadership offsite (long, dense) (Executive Summary)
“The summary contains a moderate factual distortion regarding the packaging discussion, incorrectly stating bundling protects revenue instead of add-ons.”
“The summary is flawlessly faithful to the transcript. It correctly captures all firm outcomes, explicitly records what was not decided (pricing, CS hire, packaging), uses the correct facts ($24k annual, unassigned dataset cleanup, Meridian expansion blocked), and ignores the sarcastic ECG tangent. The structure is clean and outcome-focused.”
Frequently asked
Is GPT-5 Mini better than Kimi K2.6?
Across 22 task areas we benchmarked, Kimi K2.6 ranks higher in 20 and GPT-5 Mini in 2.
Which is cheaper, GPT-5 Mini or Kimi K2.6?
GPT-5 Mini costs 1.8× less per token ($2.25 vs $4.07 per 1M).
What is GPT-5 Mini better at?
GPT-5 Mini out-ranks Kimi K2.6 at Product & Project Management, Legal & HR.
What is Kimi K2.6 better at?
Kimi K2.6 out-ranks GPT-5 Mini at Customer Support, Executive Assistant, Sales.
More comparisons
This page is Spring Prompt, running in public
We just did this for every model. Do it for your prompt.
The rankings above come from running real tasks through real models and scoring every output. Spring Prompt is that same engine — pointed at your prompt, your test cases, and your definition of good.
- Generate test cases from your prompt — no eval set required to start.
- Compare models side by side with quality, cost and latency in one matrix.
- Optimise the winner until the scores say it's ready to ship.
Prompt × model results
12 test cases · 3 evals