Benchmark Almanac

PlotPoints Round 3: Multi-Turn Arena

MiniMax-M3 1,623 leads since June 2026
Not in model cards
Version 3

Models play multi-turn roleplay scenes. Readers compare transcripts in separate standard and adult tracks. The ratings start from model judges’ assessments of writing quality.

The maker’s leaderboard ranks 21 models from 3,215 standard votes. The adult track ranks 40 models from 2,171 votes.

Released
June 2026
Kind
Creative writing
Level
Everyday
Scored
Ranked by votes
Answers
Open-ended
Tasks
All public
Measures
Preference
Language
English
In and out
Text → Text
Interaction
Multi-turn
Items from
Live play or votes
Run by
A host
Strongest model tested
GPT-5.5 #21 on the capability index
Last model added
June 2026
Scores refreshed
November 20th, 2026
40
Entries on its own leaderboard, 30 models we track

MiniMax-M3 leads with 1,623 among models released by June 2026 MiniMax-M3 leads with 1,623 among models released by June 2026

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in June 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

Bars start at 1,400.

  1. 1 MiniMax-M3 standard 1,623
  2. 2 Gemini 3.5 Flash standard 1,619
  3. 3 GPT-5.5 standard 1,593
  4. 4 Gemma 4 31B IT standard 1,592
  5. 5 Claude Sonnet 4.6 standard 1,585
  6. 6 MiMo 2.5 Pro standard 1,583
  7. 7 Qwen3.7-Max nsfw 1,573
  8. 8 Qwen3.6 27B standard 1,572
  9. 9 Claude Opus 4.8 standard 1,571
  10. 10 Gemini 3.1 Pro nsfw 1,566
  11. 11 Gemma 4 26B A4B nsfw 1,565
  12. 12 GLM-4.7 nsfw 1,563
  13. 13 Owl Alpha standard 1,543
  14. 14 GPT-4.1 nsfw 1,543
  15. 15 Claude Sonnet 4.5 nsfw 1,540
  16. 16 Magnum v4 72B nsfw 1,534
  17. 17 Gemini 3.1 Flash-Lite nsfw 1,532
  18. 18 Qwen 3.6 35B-A3B standard 1,529
  19. 19 Lunaris 8B nsfw 1,523
  20. 20 Claude Opus 4.6 nsfw 1,522
  21. 21 Mistral Small 2603 standard 1,518
  22. 22 Kimi K2.6 nsfw 1,518
  23. 23 DeepSeek-R1 May 2025, nsfw 1,517
  24. 24 Claude Opus 4.7 nsfw 1,513
  25. 25 Cydonia 24B nsfw 1,512
  26. 26 DeepSeek-V4-Flash nsfw 1,510
  27. 27 Rocinante 12B nsfw 1,508
  28. 28 DeepSeek-V3 Mar 2025, nsfw 1,508
  29. 29 Kimi K2.5 nsfw 1,507
  30. 30 Gemini 2.5 Flash Sep 2025, nsfw 1,501
  31. 31 DeepSeek-V3.2 nsfw 1,490
  32. 32 GLM-5.1 nsfw 1,480
  33. 33 Grok 4.3 nsfw 1,479
  34. 34 DeepSeek-V4-Pro nsfw 1,478
  35. 35 UnslopNemo 12B nsfw 1,476
  36. 36 Llama 4 Maverick nsfw 1,474
  37. 37 Qwen3.5-Flash nsfw 1,472
  38. 38 MiniMax-M2.7 nsfw 1,465
  39. 39 Skyfall 36B nsfw 1,453
  40. 40 Euryale L3.3 70B nsfw 1,419

Source: Official PlotPoints Round 3: Multi-Turn Arena leaderboard

Source: Official PlotPoints Round 3: Multi-Turn Arena leaderboard

#EntryScoreSettingDateSource
1 MiniMax-M3 1,623 standard Jun 2026 Official leaderboard
2 Gemini 3.5 Flash 1,619 standard May 2026 Official leaderboard
3 GPT-5.5 1,593 standard Apr 2026 Official leaderboard
4 Gemma 4 31B IT 1,592 standard Apr 2026 Official leaderboard
5 Claude Sonnet 4.6 1,585 standard Feb 2026 Official leaderboard
6 MiMo 2.5 Pro 1,583 standard Official leaderboard
7 Qwen3.7-Max 1,573 nsfw May 2026 Official leaderboard
8 Qwen3.6 27B 1,572 standard Apr 2026 Official leaderboard
9 Claude Opus 4.8 1,571 standard May 2026 Official leaderboard
10 Gemini 3.1 Pro 1,566 nsfw Feb 2026 Official leaderboard
11 Gemma 4 26B A4B 1,565 nsfw Apr 2026 Official leaderboard
12 GLM-4.7 1,563 nsfw Dec 2025 Official leaderboard
13 Owl Alpha 1,543 standard Official leaderboard
14 GPT-4.1 1,543 nsfw Apr 2025 Official leaderboard
15 Claude Sonnet 4.5 1,540 nsfw Sep 2025 Official leaderboard
16 Magnum v4 72B 1,534 nsfw Official leaderboard
17 Gemini 3.1 Flash-Lite 1,532 nsfw Mar 2026 Official leaderboard
18 Qwen 3.6 35B-A3B 1,529 standard Apr 2026 Official leaderboard
19 Lunaris 8B 1,523 nsfw Official leaderboard
20 Claude Opus 4.6 1,522 nsfw Feb 2026 Official leaderboard
21 Mistral Small 2603 1,518 standard Official leaderboard
22 Kimi K2.6 1,518 nsfw Apr 2026 Official leaderboard
23 DeepSeek-R1 May 2025 1,517 nsfw May 2025 Official leaderboard
24 Claude Opus 4.7 1,513 nsfw Apr 2026 Official leaderboard
25 Cydonia 24B 1,512 nsfw Official leaderboard
26 DeepSeek-V4-Flash 1,510 nsfw Apr 2026 Official leaderboard
27 Rocinante 12B 1,508 nsfw Official leaderboard
28 DeepSeek-V3 Mar 2025 1,508 nsfw Mar 2025 Official leaderboard
29 Kimi K2.5 1,507 nsfw Jan 2026 Official leaderboard
30 Gemini 2.5 Flash Sep 2025 1,501 nsfw Sep 2025 Official leaderboard
31 DeepSeek-V3.2 1,490 nsfw Dec 2025 Official leaderboard
32 GLM-5.1 1,480 nsfw Apr 2026 Official leaderboard
33 Grok 4.3 1,479 nsfw Apr 2026 Official leaderboard
34 DeepSeek-V4-Pro 1,478 nsfw Apr 2026 Official leaderboard
35 UnslopNemo 12B 1,476 nsfw Official leaderboard
36 Llama 4 Maverick 1,474 nsfw Apr 2025 Official leaderboard
37 Qwen3.5-Flash 1,472 nsfw Feb 2026 Official leaderboard
38 MiniMax-M2.7 1,465 nsfw Mar 2026 Official leaderboard
39 Skyfall 36B 1,453 nsfw Official leaderboard
40 Euryale L3.3 70B 1,419 nsfw Official leaderboard
#ModelScoreSettingDateSource
1 MiniMax-M3 1,623 standard Jun 2026 Official leaderboard
2 Gemini 3.5 Flash 1,619 standard May 2026 Official leaderboard
3 GPT-5.5 1,593 standard Apr 2026 Official leaderboard
4 Gemma 4 31B IT 1,592 standard Apr 2026 Official leaderboard
5 Claude Sonnet 4.6 1,585 standard Feb 2026 Official leaderboard
6 Qwen3.7-Max 1,573 nsfw May 2026 Official leaderboard
7 Qwen3.6 27B 1,572 standard Apr 2026 Official leaderboard
8 Claude Opus 4.8 1,571 standard May 2026 Official leaderboard
9 Gemini 3.1 Pro 1,566 nsfw Feb 2026 Official leaderboard
10 Gemma 4 26B A4B 1,565 nsfw Apr 2026 Official leaderboard
11 GLM-4.7 1,563 nsfw Dec 2025 Official leaderboard
12 GPT-4.1 1,543 nsfw Apr 2025 Official leaderboard
13 Claude Sonnet 4.5 1,540 nsfw Sep 2025 Official leaderboard
14 Gemini 3.1 Flash-Lite 1,532 nsfw Mar 2026 Official leaderboard
15 Qwen 3.6 35B-A3B 1,529 standard Apr 2026 Official leaderboard
16 Claude Opus 4.6 1,522 nsfw Feb 2026 Official leaderboard
17 Kimi K2.6 1,518 nsfw Apr 2026 Official leaderboard
18 DeepSeek-R1 May 2025 1,517 nsfw May 2025 Official leaderboard
19 Claude Opus 4.7 1,513 nsfw Apr 2026 Official leaderboard
20 DeepSeek-V4-Flash 1,510 nsfw Apr 2026 Official leaderboard
21 DeepSeek-V3 Mar 2025 1,508 nsfw Mar 2025 Official leaderboard
22 Kimi K2.5 1,507 nsfw Jan 2026 Official leaderboard
23 Gemini 2.5 Flash Sep 2025 1,501 nsfw Sep 2025 Official leaderboard
24 DeepSeek-V3.2 1,490 nsfw Dec 2025 Official leaderboard
25 GLM-5.1 1,480 nsfw Apr 2026 Official leaderboard
26 Grok 4.3 1,479 nsfw Apr 2026 Official leaderboard
27 DeepSeek-V4-Pro 1,478 nsfw Apr 2026 Official leaderboard
28 Llama 4 Maverick 1,474 nsfw Apr 2025 Official leaderboard
29 Qwen3.5-Flash 1,472 nsfw Feb 2026 Official leaderboard
30 MiniMax-M2.7 1,465 nsfw Mar 2026 Official leaderboard

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
MiniMax-M3
nsfw standard
1,623
Gemini 3.5 Flash
nsfw standard
1,619
GPT-5.5
nsfw standard
1,593
Gemma 4 31B IT
nsfw standard
1,592
Claude Sonnet 4.6
nsfw standard
1,585
Qwen3.7-Max
standard
1,573
Qwen3.6 27B
nsfw standard
1,572
Claude Opus 4.8
nsfw standard
1,571
Qwen 3.6 35B-A3B
nsfw standard
1,529
DeepSeek-V3
standard nsfw
1,508
Grok 4.3
standard nsfw
1,479

Source: Official PlotPoints Round 3: Multi-Turn Arena leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
MiniMax-M3 standard 1,623
Gemini 3.5 Flash standard 1,619
GPT-5.5 standard 1,593
Gemma 4 31B IT standard 1,592
Claude Sonnet 4.6 standard 1,585
Qwen3.7-Max nsfw 1,573
Qwen3.6 27B standard 1,572
Claude Opus 4.8 standard 1,571
Qwen3.7-Max standard 1,569
Gemini 3.1 Pro nsfw 1,566
Gemma 4 26B A4B nsfw 1,565
GLM-4.7 nsfw 1,563
GPT-4.1 nsfw 1,543
Claude Sonnet 4.5 nsfw 1,540
Gemini 3.1 Flash-Lite nsfw 1,532
Qwen 3.6 35B-A3B standard 1,529
Claude Opus 4.6 nsfw 1,522
Kimi K2.6 nsfw 1,518
DeepSeek-R1 May 2025 nsfw 1,517
Claude Opus 4.7 nsfw 1,513
Qwen3.6 27B nsfw 1,510
DeepSeek-V4-Flash nsfw 1,510
Gemma 4 31B IT nsfw 1,509
DeepSeek-V3 Mar 2025 nsfw 1,508
Kimi K2.5 nsfw 1,507
Gemini 2.5 Flash Sep 2025 nsfw 1,501
GPT-5.5 nsfw 1,496
Gemini 3.5 Flash nsfw 1,496
DeepSeek-V3.2 nsfw 1,490
GLM-5.1 nsfw 1,480
MiniMax-M3 nsfw 1,480
Grok 4.3 nsfw 1,479
DeepSeek-V4-Pro nsfw 1,478
DeepSeek-V3 Mar 2025 standard 1,476
Llama 4 Maverick nsfw 1,474
Qwen3.5-Flash nsfw 1,472
Claude Opus 4.8 nsfw 1,469
MiniMax-M2.7 nsfw 1,465
Qwen 3.6 35B-A3B nsfw 1,462
Claude Sonnet 4.6 nsfw 1,452
Grok 4.3 standard 1,420

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
MiniMax-M3
nsfw standard
1,623
Gemini 3.5 Flash
nsfw standard
1,619
GPT-5.5
nsfw standard
1,593
Gemma 4 31B IT
nsfw standard
1,592
Claude Sonnet 4.6
nsfw standard
1,585
Qwen3.7-Max
standard
1,573
Qwen3.6 27B
nsfw standard
1,572
Claude Opus 4.8
nsfw standard
1,571
Qwen 3.6 35B-A3B
nsfw standard
1,529
DeepSeek-V3
standard nsfw
1,508
Grok 4.3
standard nsfw
1,479

Source: Official PlotPoints Round 3: Multi-Turn Arena leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
MiniMax-M3 standard 1,623
Gemini 3.5 Flash standard 1,619
GPT-5.5 standard 1,593
Gemma 4 31B IT standard 1,592
Claude Sonnet 4.6 standard 1,585
Qwen3.7-Max nsfw 1,573
Qwen3.6 27B standard 1,572
Claude Opus 4.8 standard 1,571
Qwen3.7-Max standard 1,569
Gemini 3.1 Pro nsfw 1,566
Gemma 4 26B A4B nsfw 1,565
GLM-4.7 nsfw 1,563
GPT-4.1 nsfw 1,543
Claude Sonnet 4.5 nsfw 1,540
Gemini 3.1 Flash-Lite nsfw 1,532
Qwen 3.6 35B-A3B standard 1,529
Claude Opus 4.6 nsfw 1,522
Kimi K2.6 nsfw 1,518
DeepSeek-R1 May 2025 nsfw 1,517
Claude Opus 4.7 nsfw 1,513
Qwen3.6 27B nsfw 1,510
DeepSeek-V4-Flash nsfw 1,510
Gemma 4 31B IT nsfw 1,509
DeepSeek-V3 Mar 2025 nsfw 1,508
Kimi K2.5 nsfw 1,507
Gemini 2.5 Flash Sep 2025 nsfw 1,501
GPT-5.5 nsfw 1,496
Gemini 3.5 Flash nsfw 1,496
DeepSeek-V3.2 nsfw 1,490
GLM-5.1 nsfw 1,480
MiniMax-M3 nsfw 1,480
Grok 4.3 nsfw 1,479
DeepSeek-V4-Pro nsfw 1,478
DeepSeek-V3 Mar 2025 standard 1,476
Llama 4 Maverick nsfw 1,474
Qwen3.5-Flash nsfw 1,472
Claude Opus 4.8 nsfw 1,469
MiniMax-M2.7 nsfw 1,465
Qwen 3.6 35B-A3B nsfw 1,462
Claude Sonnet 4.6 nsfw 1,452
Grok 4.3 standard 1,420

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

MiniMax
Own leaderboard Not tested since Jun 2026
MiniMax-M3 1,623

Source: Official PlotPoints Round 3: Multi-Turn Arena leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard
1,400 1,450 1,500 1,550 1,600 1,650 2025 2026 1,400 1,450 1,500 1,550 1,600 1,650 2025 2026

Source: Official PlotPoints Round 3: Multi-Turn Arena leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

Versions

More in Writing and taste

All 65 in Writing and taste
  • Design Bench (Moonshot) In use Models make graphic designs from direct prompts or few-shot reference images. An automated grader rates the results as Elo scores. Kimi K3.1 Pro 957
  • WritingBench Fading Models complete 1,000 real-world writing requests across 6 domains and 100 subdomains. A judge model scores each reply on five query-specific criteria. GPT-5 83.9
  • Creative Writing (MiniMax in-house) Only in system cards Models write creative pieces for real Hailuo AI user requests. Automatic assessments score the writing. MiniMax-Text-01 81.3%
  • DeepSeek Chinese functional writing evaluation Only in system cards Models answer 3,170 everyday Chinese writing prompts such as reports, emails and speeches. Pairwise judging against a baseline model gives a win rate. DeepSeek-V4-Pro 62.6%
  • Slide generation human evaluation (Z.ai) Only in system cards People compare HTML slide decks from two models on content, layout and visual quality. Metric is overall win rate. GLM-5 67.5%
  • Task-dependent output homogenization evaluation Only in system cards Models answer prompts several times. A judge scores functional diversity where it is wanted and consistency where one answer is right. Muse Spark 0.21

Data audit

All 9 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 3 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (40 entries).
  • Pass Models scored 30 models we track, plus 10 other entries on its leaderboard.
  • Pass Newest score The newest is from June 2026, for MiniMax-M3.
  • Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GPT-5.5 (#21).
  • Note Headroom Scored as an Elo rating, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on plotlightstudios.com: 41 rows for models we track and 10 other entries at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check