Undominated.ai

Living stamp

Now

A dated stamp of the frontier, not a price index. No chained deflation number. The archive started when we started measuring; we do not backfill.

Priced for the balanced mix: three parts input to one part output, no cache assumed. Changing the mix is a different question; this page is a stamp, not a slider.

As of 2026-10-06 2026-10-06

12 12 undominated

models 364
providers 55
rated, priced 145
undominated 12
strictly beaten 92%
133 of 145 rated, priced models have another that scores at least as high, costs no more, and wins on one of the two. The short chartreuse length is the undominated set.
$0.030$0.100$0.300$1.00$3.00$10.00$30.0010801170126013501440Claude Opus 5.5 · 1511.7 · $8.00/MGemini 3.8 Flash · 1497.0 · $3.00/MMiMo-V2.6-Pro · 1491.0 · $0.540/MGLM 5.3 Flash · 1469.6 · $0.238/MDeepSeek V4.1 Flash · 1462.5 · $0.182/MMiMo-V2.6-Flash · 1456.4 · $0.175/MGemma 4 26B A4B · 1433.9 · $0.127/MDeepSeek V4 Flash 0423 · 1432.1 · $0.113/Mgpt-oss-120b · 1365.4 · $0.065/MGemma 3 4B · 1290.8 · $0.063/Mgpt-oss-20b · 1287.7 · $0.036/MLlama 3.1 8B Instruct · 1186.5 · $0.025/MLMArenaeffective $/M — log
A staircase, not a curve. Between two rungs there is no model, so a smooth line would invent options that do not exist. Chartreuse marks the undominated set. Each graphite dot has a model that scores at least as high, costs no more, and wins on one of the two. Logos name the vendor; they are not a ranking colour.

The rungs, best first

  1. 1 Claude Opus 5.5 Anthropic
    LMArena 1511.7
    $/M $8.00/M

    Check vs Gemini 3.8 Flash

  2. 2 Gemini 3.8 Flash Google
    LMArena 1497.0
    $/M $3.00/M

    Check vs MiMo-V2.6-Pro

  3. 3 MiMo-V2.6-Pro Xiaomi
    LMArena 1491.0
    $/M $0.540/M

    Check vs GLM 5.3 Flash

  4. 4 GLM 5.3 Flash Z.ai
    LMArena 1469.6
    $/M $0.238/M

    Check vs DeepSeek V4.1 Flash

  5. 5 DeepSeek V4.1 Flash DeepSeek
    LMArena 1462.5
    $/M $0.182/M

    Check vs MiMo-V2.6-Flash

  6. 6 MiMo-V2.6-Flash Xiaomi
    LMArena 1456.4
    $/M $0.175/M

    Check vs Gemma 4 26B A4B

  7. 7 Gemma 4 26B A4B Google
    LMArena 1433.9
    $/M $0.127/M

    Check vs DeepSeek V4 Flash 0423

  8. 8 DeepSeek V4 Flash 0423 DeepSeek
    LMArena 1432.1
    $/M $0.113/M

    Check vs gpt-oss-120b

  9. 9 gpt-oss-120b OpenAI
    LMArena 1365.4
    $/M $0.065/M

    Check vs Gemma 3 4B

  10. 10 Gemma 3 4B Google
    LMArena 1290.8
    $/M $0.063/M

    Check vs gpt-oss-20b

  11. 11 gpt-oss-20b OpenAI
    LMArena 1287.7
    $/M $0.036/M

    Check vs Llama 3.1 8B Instruct

  12. 12 Llama 3.1 8B Instruct Meta
    LMArena 1186.5
    $/M $0.025/M

    Check

Who is on it

  • Google 3 of 12
  • DeepSeek 2 of 12
  • OpenAI 2 of 12
  • Xiaomi 2 of 12
  • Anthropic 1 of 12
  • Meta 1 of 12
  • Z.ai 1 of 12

Stamps we have

The dated archive starts with the first snapshot. There is no invented history.

  1. 2026-08-24 10 undominated
  2. 2026-08-25 10 undominated
  3. 2026-08-26 9 undominated
  4. 2026-08-28 9 undominated
  5. 2026-08-29 9 undominated
  6. 2026-08-30 9 undominated
  7. 2026-09-02 9 undominated
  8. 2026-09-03 10 undominated
  9. 2026-09-04 10 undominated
  10. 2026-09-08 10 undominated
  11. 2026-09-11 10 undominated
  12. 2026-09-15 11 undominated
  13. 2026-09-22 12 undominated
  14. 2026-09-23 13 undominated
  15. 2026-09-29 12 undominated
  16. 2026-09-30 12 undominated
  17. 2026-10-06 12 undominated

Cite

APA 
Undominated.ai. (2026-10-06). The value frontier. https://undominated.ai/now/
BibTeX 
@misc{undominated-now,
  author = {{Undominated.ai}},
  title  = {The value frontier},
  year   = {2026},
  url    = {https://undominated.ai/now/},
  note   = {2026-10-06}
}

LMArena scores used under CC BY 4.0

Rank projection 06cd12db54e2968c3de49b45a3a2a6881f374f6e58a462437784450842efb989

Open the live frontier plot Open the live board

Price at pinned floors

Table until eight weekly stamps.

Price at pinned LMArena Elo floors, standard delivery, as-of dates on the columns. No CPI.
Floor
lmarena-1200-v1 ≥1200upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053openai/gpt-oss-20b $0.055openai/gpt-oss-20b $0.055deepseek/deepseek-v4-flash $0.048openai/gpt-oss-20b $0.055openai/gpt-oss-20b $0.036openai/gpt-oss-20b $0.036openai/gpt-oss-20b $0.036openai/gpt-oss-20b $0.036
lmarena-1350-v1 ≥1350upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053upstage/solar-pro4 $0.053openai/gpt-oss-120b $0.070openai/gpt-oss-120b $0.070deepseek/deepseek-v4-flash $0.048qwen/qwen3-30b-a3b-instruct-2507 $0.084qwen/qwen3-30b-a3b-instruct-2507 $0.084openai/gpt-oss-120b $0.070openai/gpt-oss-120b $0.070openai/gpt-oss-120b $0.065
lmarena-1400-v1 ≥1400deepseek/deepseek-v4-flash $0.061deepseek/deepseek-v4-flash $0.061deepseek/deepseek-v4-flash $0.101deepseek/deepseek-v4-flash $0.101deepseek/deepseek-v4-flash $0.101deepseek/deepseek-v4-flash $0.101deepseek/deepseek-v4-flash $0.111deepseek/deepseek-v4-flash $0.099deepseek/deepseek-v4-flash $0.110deepseek/deepseek-v4-flash $0.107google/gemma-4-26b-a4b-it $0.087deepseek/deepseek-v4-flash $0.109deepseek/deepseek-v4-flash $0.048deepseek/deepseek-v4-flash $0.111deepseek/deepseek-v4-flash $0.103deepseek/deepseek-v4-flash $0.095google/gemma-4-26b-a4b-it $0.121deepseek/deepseek-v4-flash $0.113
lmarena-1450-v1 ≥1450deepseek/deepseek-v4-pro $0.496deepseek/deepseek-v4-pro $0.496xiaomi/mimo-v2.5-pro $0.544xiaomi/mimo-v2.5-pro $0.544xiaomi/mimo-v2.5-pro $0.544xiaomi/mimo-v2.5-pro $0.544xiaomi/mimo-v2.5-pro $0.544xiaomi/mimo-v2.5-pro $0.544xiaomi/mimo-v2.5-pro $0.544xiaomi/mimo-v2.5-pro $0.544z-ai/glm-5.3-flash $0.238z-ai/glm-5.3-flash $0.238z-ai/glm-5.3-flash $0.143z-ai/glm-5.3-flash $0.238z-ai/glm-5.3-flash $0.238z-ai/glm-5.3-flash $0.238z-ai/glm-5.3-flash $0.238xiaomi/mimo-v2.6-flash $0.175
Evidence & Ask
Markdown for LLMs