Part of Best LLM for Summarization

Best LLM for Document Summarization

Gemini 4 Carbon ranks first for Document Summarization as of November 15, 2026.[1][2] It ranks #1 of 148 on LMArena's long-query category (Elo 1549), based on blind human preference for longer prompts. Gemini 3.7 Flash is the next-ranked alternative. It scores 79.93% on LiveBench Instruction Following (#2 of 59), including paraphrasing, simplifying, story generation, and summarization.

About this recommendation

Updated
Evidence through
Sources
11
Revision
v130

Decision audit

Why this result

Inspect the inputs and the computed order behind the recommendation.

Models screened

24

live candidates

Evaluation feeds

6

task-weighted

Winner coverage

45%

intended feed weight

Largest provider share

5 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 43%.

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.

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
FACTS GroundingObserved 2026-10-29 · evaluated 2026-10-29 · Gemini 4 Carbon
38%
#613/24
LiveBench Instruction Followingfailing; last successful check 2026-10-18
31%
not measured20/24
LMArena Documentstale
0%
no current result0/24
LongBench v2disabled; last successful check unknownunavailable
0%
feed unavailable0/24
LMArena Long QueryObserved 2026-11-11 · evaluation date not supplied · high effort
15%
#124/24
LMArena Text FactualityObserved 2026-11-11 · evaluation date not supplied · high effort
15%
#124/24

Provider concentration

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

Google63%
  • Google5 models
  • OpenAI2 models
  • 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 45%

    #1 LMArena Long Query · #1 LMArena Text Factuality

  2. Google · coverage 40%

    #2 LiveBench Instruction Following · #13 LMArena Long Query

  3. Google · coverage 45%

    #2 LMArena Long Query · #4 LMArena Text Factuality

  4. Meta · coverage 40%

    #4 LiveBench Instruction Following · #13 LMArena Text Factuality

  5. Google · coverage 65%

    #1 LiveBench Instruction Following · #9 LMArena Long Query

  6. OpenAI · coverage 65%

    #3 FACTS Grounding · #7 LiveBench Instruction Following

  7. OpenAI · coverage 65%

    #4 FACTS Grounding · #11 LiveBench Instruction Following

  8. Google · coverage 40%

    #6 LiveBench Instruction Following · #15 LMArena Text Factuality

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 long-query category (Elo 1549), based on blind human preference for longer prompts.

    Best when: Its strongest result for this task is on LMArena Long Query, which measures handling long, detailed prompts.
    Source 1
    “Within the 148 catalog models matched to this evaluation: Ranks #1 of 148 on LMArena's long-query category (Elo 1549), based on blind human preference for longer prompts.”
    LMArena long-query categoryOpen original ↗

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

  2. 02

    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 2
    “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.

  3. 03

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

    Ranks #2 on LMArena's long-query category (Elo 1547) and overall text arena (Elo 1522), with solid factuality at #4 of 112.

    Best when: Its strongest result for this task is on LMArena Long Query, which measures handling long, detailed prompts.
    Source 3
    “Within the 148 catalog models matched to this evaluation: Ranks #2 of 148 on LMArena's long-query category (Elo 1547), based on blind human preference for longer prompts.”
    LMArena long-query categoryOpen original ↗

    Watch out for

    • Verify source alignment carefully; it scores 73.4% (#9 of 25) on FACTS Grounding, below top performers.
      Source 4
      “Within the 25 catalog models matched to this evaluation: Scores 73.4% on FACTS Grounding (#9 of 25), measuring factual support against supplied source material.”
      FACTS GroundingOpen original ↗
  4. 04

    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 5
    “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 ↗

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

  5. 05

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

    Leads LiveBench Instruction Following at #1 (81.41%) and ranks #7 on document arena, though factuality lags at #25.

    Best when: Its strongest result for this task is on LiveBench Instruction Following, which measures following detailed instructions exactly.
    Source 6
    “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 ↗

    Watch out for

    • Verify factual claims independently; it ranks #25 of 112 on LMArena's factuality text view.
      Source 7
      “Within the 112 catalog models matched to this evaluation: Ranks #25 of 112 on LMArena's factuality text view (Elo 1470), where answers are judged for factual accuracy in blind comparison.”
      LMArena text factualityOpen original ↗
    • Check grounding against sources; it scores 68.2% (#11 of 25) on FACTS Grounding.
      Source 8
      “Within the 25 catalog models matched to this evaluation: Scores 68.2% on FACTS Grounding (#11 of 25), measuring factual support against supplied source material.”
      FACTS GroundingOpen original ↗
  6. 06

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

    GPT-6 Astra scores 80.1% on FACTS Grounding (#3 of 25), measuring factual support against supplied source material.

    Best when: Its strongest result for this task is on FACTS Grounding, which measures staying faithful to supplied source documents.
    Source 9
    “Within the 25 catalog models matched to this evaluation: Scores 80.1% on FACTS Grounding (#3 of 25), measuring factual support against supplied source material.”
    FACTS GroundingOpen original ↗

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

  7. 07

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

    GPT-6.1 Sol scores 78% on FACTS Grounding (#4 of 25), measuring factual support against supplied source material.

    Best when: Its strongest result for this task is on FACTS Grounding, which measures staying faithful to supplied source documents.
    Source 10
    “Within the 25 catalog models matched to this evaluation: Scores 78% on FACTS Grounding (#4 of 25), measuring factual support against supplied source material.”
    FACTS GroundingOpen original ↗

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

  8. 08

    Gemini 3.5 FlashGoogle · $1.50 in / $9 out per M tokens · 1.0M context

    Gemini 3.5 Flash scores 75.6% on LiveBench Instruction Following (#6 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 11
    “Within the 59 catalog models matched to this evaluation: Scores 75.6% on LiveBench Instruction Following (#6 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.

Frequently asked

What is the best model for Document Summarization?
Gemini 4 Carbon ranks first as of November 15, 2026. Gemini 4 Carbon ranks #1 of 148 on LMArena's long-query category (Elo 1549), based on blind human preference for longer prompts.[1]
What is the best alternative to Gemini 4 Carbon?
Gemini 3.7 Flash ranks second. Gemini 3.7 Flash scores 79.93% on LiveBench Instruction Following (#2 of 59), including paraphrasing, simplifying, story generation, and summarization.[2]

Sources

  1. 1

    “Within the 148 catalog models matched to this evaluation: Ranks #1 of 148 on LMArena's long-query category (Elo 1549), based on blind human preference for longer prompts.”

    LMArena long-query category · Benchmark · Nov 11, 2026
  2. 2

    “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
  3. 3

    “Within the 148 catalog models matched to this evaluation: Ranks #2 of 148 on LMArena's long-query category (Elo 1547), based on blind human preference for longer prompts.”

    LMArena long-query category · Benchmark · Nov 11, 2026
  4. 4

    “Within the 25 catalog models matched to this evaluation: Scores 73.4% on FACTS Grounding (#9 of 25), measuring factual support against supplied source material.”

    FACTS Grounding · Benchmark · Oct 24, 2026
  5. 5

    “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
  6. 6

    “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
  7. 7

    “Within the 112 catalog models matched to this evaluation: Ranks #25 of 112 on LMArena's factuality text view (Elo 1470), where answers are judged for factual accuracy in blind comparison.”

    LMArena text factuality · Benchmark · Nov 11, 2026
  8. 8

    “Within the 25 catalog models matched to this evaluation: Scores 68.2% on FACTS Grounding (#11 of 25), measuring factual support against supplied source material.”

    FACTS Grounding · Benchmark · Oct 23, 2026
  9. 9

    “Within the 25 catalog models matched to this evaluation: Scores 80.1% on FACTS Grounding (#3 of 25), measuring factual support against supplied source material.”

    FACTS Grounding · Benchmark · Oct 24, 2026
  10. 10

    “Within the 25 catalog models matched to this evaluation: Scores 78% on FACTS Grounding (#4 of 25), measuring factual support against supplied source material.”

    FACTS Grounding · Benchmark · Oct 24, 2026
  11. 11

    “Within the 59 catalog models matched to this evaluation: Scores 75.6% on LiveBench Instruction Following (#6 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.