Part of Best LLM for Math & Reasoning

Best LLM for Logic & Problem Solving

Gemini 4 Carbon ranks first for Logic & Problem Solving as of November 15, 2026.[1][2] It ranks #1 of 148 on LMArena's instruction-following category (Elo 1539), based on blind human preference votes. Gemini 4 Argon scores within two points, so read the top two as close rather than as a measured gap. Gemini 4 Argon is the next-ranked alternative. It ranks #1 of 143 on LMArena's maths category (Elo 1544), based on blind human preference votes for maths prompts.

About this recommendation

Updated
Evidence through
Sources
10
Revision
v134

Decision audit

Why this result

Inspect the inputs and the computed order behind the recommendation.

Models screened

24

live candidates

Evaluation feeds

5

task-weighted

Winner coverage

41%

intended feed weight

Largest provider share

4 of 8

Google

Established source breadth. 3 citation families and 0 practitioner families provide attributed capability reports; 0 cautionary threads retained. The largest citation family contributes 50%.

Google: Gemini 4 Carbon leads the measured task comparison with 2 independent evaluation families. Community volume, release age and usage popularity add no ranking points. The leading scores are close under this policy.

Real-world track record is still developing. The recommendation can lead on independent measurements while practitioner reports accumulate.

Sources evaluated

These are the weights used in scoring. Missing measurements are omitted from each model’s average. A feed with no result for any model released in the last 60 days is stale and does not score until it catches up.

winner: Gemini 4 Carbon
Evaluation feedWeightWinner resultField measured
LiveBench Reasoningfailing; last successful check 2026-10-18
32%
not measured16/24
LMArena Instruction FollowingObserved 2026-11-11 · evaluation date not supplied · high effort
23%
#124/24
GPQA Diamond (Epoch AI)Observed 2026-11-02 · evaluated 2026-11-02 · high effort · standard error 1.20
18%
#122/24
LiveBench Instruction Followingfailing; last successful check 2026-10-18
14%
not measured16/24
LMArena Math
14%
not measured23/24

Provider concentration

Each exact model is scored separately; provider identity is not a ranking input.

Google50%
  • Google4 models
  • OpenAI2 models
  • Anthropic1 model
  • Meta1 model

Decision table

Every published model is shown in computed order. Practitioner sources are distinct community threads, not the citations repeated in the prose below.

  1. Google · coverage 41%

    #1 GPQA Diamond (Epoch AI) · #1 LMArena Instruction Following

  2. Google · coverage 55% · close score

    #1 LMArena Math · #2 GPQA Diamond (Epoch AI)

  3. Google · coverage 100%

    #1 LiveBench Instruction Following · #4 LMArena Math

  4. OpenAI · coverage 100%

    #1 LiveBench Reasoning · #3 GPQA Diamond (Epoch AI)

  5. OpenAI · coverage 100%

    #2 LiveBench Reasoning · #5 GPQA Diamond (Epoch AI)

  6. Google · coverage 100%

    #2 LiveBench Instruction Following · #7 GPQA Diamond (Epoch AI)

  7. Anthropic · coverage 100%

    #3 LiveBench Reasoning · #3 LMArena Instruction Following

  8. Meta · coverage 82%

    #4 LiveBench Instruction Following · #12 LiveBench Reasoning

Task score averages the available, task-weighted measurements. Benchmark positions compare models matched to our catalog. Models measured by at least two independent evaluator families are considered first. Missing tests, practitioner report counts, popularity and release age add no points. Coverage describes evidence breadth, not a probability of correctness. Scores within two points are shown as close under this policy; this is not a statistical confidence interval. When the leading scores are that close, the model measured by more evaluator families ranks first.

  1. 01

    Gemini 4 CarbonGoogle · $2 in / $10 out per M tokens · 1.0M context

    Gemini 4 Carbon ranks #1 of 148 on LMArena's instruction-following category (Elo 1539), based on blind human preference votes.

    Best when: Its strongest result for this task is on LMArena Instruction Following, which measures following instructions as people judge them.
    Source 1
    “Within the 148 catalog models matched to this evaluation: Ranks #1 of 148 on LMArena's instruction-following category (Elo 1539), based on blind human preference votes.”
    LMArena instruction-following categoryOpen original ↗

    No practitioner reports are linked yet, so no cautions are listed. Absence of reports is not evidence of reliability.

  2. 02

    Gemini 4 ArgonGoogle · $2 in / $10 out per M tokens · 1.0M context

    Gemini 4 Argon ranks #1 of 143 on LMArena's maths category (Elo 1544), based on blind human preference votes for maths prompts.

    Best when: Its strongest result for this task is on LMArena Math, which measures math answers people prefer in blind comparison.
    Source 2
    “Within the 143 catalog models matched to this evaluation: Ranks #1 of 143 on LMArena's maths category (Elo 1544), based on blind human preference votes for maths prompts.”
    LMArena maths categoryOpen original ↗

    No practitioner reports are linked yet, so no cautions are listed. Absence of reports is not evidence of reliability.

  3. 03

    Gemini 3.8 FlashGoogle · $0.83 in / $4.13 out per M tokens · 1.0M context

    Gemini 3.8 Flash scores 81.41% on LiveBench Instruction Following (#1 of 59), including paraphrasing, simplifying, story generation, and summarization.

    Best when: Its strongest result for this task is on LiveBench Instruction Following, which measures following detailed instructions exactly.
    Source 3
    “Within the 59 catalog models matched to this evaluation: Scores 81.41% on LiveBench Instruction Following (#1 of 59), including paraphrasing, simplifying, story generation, and summarization.”
    LiveBench Instruction FollowingOpen original ↗
  4. 04

    GPT-6 AstraOpenAI · $10 in / $50 out per M tokens · 1.1M context

    GPT-6 Astra scores 92.65% on LiveBench Reasoning (#1 of 59), an objective, ground-truth-scored evaluation refreshed with new questions.

    Best when: Its strongest result for this task is on LiveBench Reasoning, which measures multi-step reasoning.
    Source 4
    “Within the 59 catalog models matched to this evaluation: Scores 92.65% on LiveBench Reasoning (#1 of 59), an objective, ground-truth-scored evaluation refreshed with new questions.”
    LiveBench ReasoningOpen original ↗
  5. 05

    GPT-6.1 SolOpenAI · $2 in / $10 out per M tokens · 1.1M context

    Ranks #2 on LiveBench Reasoning at 92.63% and passes all 64 calibration controls in a focused evaluation of everyday scenarios without assistance.

    Best when: Use for objective reasoning tasks: scores 92.63% on LiveBench Reasoning, ranking #2 of 59.
    Source 5
    “Within the 59 catalog models matched to this evaluation: Scores 92.63% on LiveBench Reasoning (#2 of 59), an objective, ground-truth-scored evaluation refreshed with new questions.”
    LiveBench ReasoningOpen original ↗

    More tips

    • Deploy for calibration-sensitive tasks: passed all 64 calibration controls and four repaired everyday scenarios without extra questions or assistance.
      Source 6
      “Conclusion All 64 calibration controls and all four repaired everyday scenarios pass, without extra questions or assistance. Goal Verify scenarios 30, 12, 32 and 46 after their reviewed repairs, preserving the original failures in #698. Stop after this focused round for #668. Model and scope - OpenAI gpt-6.1-sol, high, standard tier, text only, one repetition. - Exactly everyday-30, everyday-12, everyday-32 and everyday-46, in that order. - All 64 judge calibration controls must pass…”

    Watch out for

    • Expect lower human preference on instruction-following: ranks #33 of 148 on LMArena's instruction-following category with Elo 1463.
      Source 7
      “Within the 148 catalog models matched to this evaluation: Ranks #33 of 148 on LMArena's instruction-following category (Elo 1463), based on blind human preference votes.”
      LMArena instruction-following categoryOpen original ↗
  6. 06

    Gemini 3.7 FlashGoogle · $0.75 in / $3.75 out per M tokens · 1.0M context

    Gemini 3.7 Flash scores 79.93% on LiveBench Instruction Following (#2 of 59), including paraphrasing, simplifying, story generation, and summarization.

    Best when: Its strongest result for this task is on LiveBench Instruction Following, which measures following detailed instructions exactly.
    Source 8
    “Within the 59 catalog models matched to this evaluation: Scores 79.93% on LiveBench Instruction Following (#2 of 59), including paraphrasing, simplifying, story generation, and summarization.”
    LiveBench Instruction FollowingOpen original ↗

    No practitioner reports are linked yet, so no cautions are listed. Absence of reports is not evidence of reliability.

  7. 07

    Claude Opus 5.5Anthropic · $4 in / $20 out per M tokens · 1M context

    Claude Opus 5.5 ranks #3 of 148 on LMArena's instruction-following category (Elo 1535), based on blind human preference votes.

    Best when: Its strongest result for this task is on LMArena Instruction Following, which measures following instructions as people judge them.
    Source 9
    “Within the 148 catalog models matched to this evaluation: Ranks #3 of 148 on LMArena's instruction-following category (Elo 1535), based on blind human preference votes.”
    LMArena instruction-following categoryOpen original ↗

    No practitioner reports are linked yet, so no cautions are listed. Absence of reports is not evidence of reliability.

  8. 08

    Muse Spark 1.3Meta · $1.25 in / $4.25 out per M tokens · 1.0M context

    Muse Spark 1.3 scores 78% on LiveBench Instruction Following (#4 of 59), including paraphrasing, simplifying, story generation, and summarization.

    Best when: Its strongest result for this task is on LiveBench Instruction Following, which measures following detailed instructions exactly.
    Source 10
    “Within the 59 catalog models matched to this evaluation: Scores 78% on LiveBench Instruction Following (#4 of 59), including paraphrasing, simplifying, story generation, and summarization.”
    LiveBench Instruction FollowingOpen original ↗

Frequently asked

What is the best model for Logic & Problem Solving?
Gemini 4 Carbon ranks first as of November 15, 2026. Gemini 4 Carbon ranks #1 of 148 on LMArena's instruction-following category (Elo 1539), based on blind human preference votes.[1]
What is the best alternative to Gemini 4 Carbon?
Gemini 4 Argon ranks second. Gemini 4 Argon ranks #1 of 143 on LMArena's maths category (Elo 1544), based on blind human preference votes for maths prompts.[2]

Sources

  1. 1

    “Within the 148 catalog models matched to this evaluation: Ranks #1 of 148 on LMArena's instruction-following category (Elo 1539), based on blind human preference votes.”

    LMArena instruction-following category · Benchmark · Nov 11, 2026
  2. 2

    “Within the 143 catalog models matched to this evaluation: Ranks #1 of 143 on LMArena's maths category (Elo 1544), based on blind human preference votes for maths prompts.”

    LMArena maths category · Benchmark · Nov 11, 2026
  3. 3

    “Within the 59 catalog models matched to this evaluation: Scores 81.41% on LiveBench Instruction Following (#1 of 59), including paraphrasing, simplifying, story generation, and summarization.”

    LiveBench Instruction Following · Benchmark · Oct 18, 2026
  4. 4

    “Within the 59 catalog models matched to this evaluation: Scores 92.65% on LiveBench Reasoning (#1 of 59), an objective, ground-truth-scored evaluation refreshed with new questions.”

    LiveBench Reasoning · Benchmark · Oct 18, 2026
  5. 5

    “Within the 59 catalog models matched to this evaluation: Scores 92.63% on LiveBench Reasoning (#2 of 59), an objective, ground-truth-scored evaluation refreshed with new questions.”

    LiveBench Reasoning · Benchmark · Oct 18, 2026
  6. 6

    “Conclusion All 64 calibration controls and all four repaired everyday scenarios pass, without extra questions or assistance. Goal Verify scenarios 30, 12, 32 and 46 after their reviewed repairs, preserving the original failures in #698. Stop after this focused round for #668. Model and scope - OpenAI gpt-6.1-sol, high, standard tier, text only, one repetition. - Exactly everyday-30, everyday-12, everyday-32 and everyday-46, in that order. - All 64 judge calibration controls must pass…”

    johlju · GitHub · Nov 6, 2026
  7. 7

    “Within the 148 catalog models matched to this evaluation: Ranks #33 of 148 on LMArena's instruction-following category (Elo 1463), based on blind human preference votes.”

    LMArena instruction-following category · Benchmark · Nov 11, 2026
  8. 8

    “Within the 59 catalog models matched to this evaluation: Scores 79.93% on LiveBench Instruction Following (#2 of 59), including paraphrasing, simplifying, story generation, and summarization.”

    LiveBench Instruction Following · Benchmark · Oct 18, 2026
  9. 9

    “Within the 148 catalog models matched to this evaluation: Ranks #3 of 148 on LMArena's instruction-following category (Elo 1535), based on blind human preference votes.”

    LMArena instruction-following category · Benchmark · Nov 11, 2026
  10. 10

    “Within the 59 catalog models matched to this evaluation: Scores 78% on LiveBench Instruction Following (#4 of 59), including paraphrasing, simplifying, story generation, and summarization.”

    LiveBench Instruction Following · Benchmark · Oct 18, 2026

Rankings synthesized from community evidence and open benchmarks. See methodology. Not driven by vendor marketing.