Benchmark Almanac

Letta Leaderboard

GPT-5.2 Codex 93.0 leads since December 2025
Not in model cards

Agents read, write and update core and archival memory while answering synthetic factual questions. A model judge scores answers and penalizes unnecessary memory operations.

Letta's launch post reports average scores and costs for the top 10 models across memory read, write and update tasks. The archived repository now points users to Letta Evals.

Made by
Letta
Released
May 2025
Kind
Learning from the context
Level
Expert
Scored
Graded by a model
Answers
Agent task
Tasks
All public
Measures
Capability
Language
English
In and out
Text → Text, Actions
Interaction
Agentic
Environment
APIs and tools
Items from
Generated
Refresh
One fixed set
Run by
A host
In views
Agents
Strongest model tested
GPT-5.4 #28 on the capability index
Last model added
March 2026
Scores refreshed
November 21st, 2026
15
Entries on its own leaderboard, 14 models we track

GPT-5.2 Codex leads with 93.0 among models released by March 2026 GPT-5.2 Codex leads with 93.0 among models released by March 2026

Best score per entry from the benchmark's own leaderboard, read on November 21st, 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 March 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-5.2 Codex xhigh 93.0
  2. 2 GPT-5.4 xhigh 89.0
  3. 3 Claude Sonnet 4.6 88.0
  4. 4 GPT-5.2 xhigh 87.0
  5. 5 GPT-5.3 Codex 85.0
  6. 6 Claude Opus 4.6 84.0
  7. 7 Gemini 3 Flash 82.0
  8. 8 google/gemini-3.1 79.0
  9. 9 GPT-5 mini high 67.0
  10. 10 Claude Opus 4.5 63.0
  11. 11 Kimi K2.5 57.0
  12. 12 Claude Sonnet 4.5 48.0
  13. 13 Claude Haiku 4.5 43.0
  14. 14 GLM-5 39.0
  15. 15 MiniMax-M2.5 37.0

Source: Official Letta Leaderboard leaderboard

#EntryScoreSettingDateSource
1 GPT-5.2 Codex 93.0 xhigh Dec 2025 Official leaderboard
2 GPT-5.4 89.0 xhigh Mar 2026 Official leaderboard
3 Claude Sonnet 4.6 88.0 Feb 2026 Official leaderboard
4 GPT-5.2 87.0 xhigh Dec 2025 Official leaderboard
5 GPT-5.3 Codex 85.0 Feb 2026 Official leaderboard
6 Claude Opus 4.6 84.0 Feb 2026 Official leaderboard
7 Gemini 3 Flash 82.0 Dec 2025 Official leaderboard
8 google/gemini-3.1 79.0 Official leaderboard
9 GPT-5 mini 67.0 high Aug 2025 Official leaderboard
10 Claude Opus 4.5 63.0 Nov 2025 Official leaderboard
11 Kimi K2.5 57.0 Jan 2026 Official leaderboard
12 Claude Sonnet 4.5 48.0 Sep 2025 Official leaderboard
13 Claude Haiku 4.5 43.0 Oct 2025 Official leaderboard
14 GLM-5 39.0 Feb 2026 Official leaderboard
15 MiniMax-M2.5 37.0 Feb 2026 Official leaderboard
#ModelScoreSettingDateSource
1 GPT-5.2 Codex 93.0 xhigh Dec 2025 Official leaderboard
2 GPT-5.4 89.0 xhigh Mar 2026 Official leaderboard
3 Claude Sonnet 4.6 88.0 Feb 2026 Official leaderboard
4 GPT-5.2 87.0 xhigh Dec 2025 Official leaderboard
5 GPT-5.3 Codex 85.0 Feb 2026 Official leaderboard
6 Claude Opus 4.6 84.0 Feb 2026 Official leaderboard
7 Gemini 3 Flash 82.0 Dec 2025 Official leaderboard
8 GPT-5 mini 67.0 high Aug 2025 Official leaderboard
9 Claude Opus 4.5 63.0 Nov 2025 Official leaderboard
10 Kimi K2.5 57.0 Jan 2026 Official leaderboard
11 Claude Sonnet 4.5 48.0 Sep 2025 Official leaderboard
12 Claude Haiku 4.5 43.0 Oct 2025 Official leaderboard
13 GLM-5 39.0 Feb 2026 Official leaderboard
14 MiniMax-M2.5 37.0 Feb 2026 Official leaderboard

Rankings by part

We read 3 parts from the benchmark’s own leaderboard: Filesystem Rubric, Task Completion Rubric and Skill Use Rubric. Each ranks models on one part of the test. GPT-5.2 Codex leads 1 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.

Filesystem Rubric

14 models we track have a score here.

  1. 1 GPT-5.2 Codex 93.0
  2. 2 GPT-5.4 89.0
  3. 3 Claude Sonnet 4.6 88.0
  4. 4 GPT-5.2 87.0
  5. 5 GPT-5.3 Codex 85.0
  6. 6 Claude Opus 4.6 84.0
  7. 7 Gemini 3 Flash 82.0
  8. 8 GPT-5 mini 67.0
  9. 9 Claude Opus 4.5 63.0
  10. 10 Kimi K2.5 57.0

The best score, 93.0, costs $44 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

OpenAI Google DeepMind Moonshot AI Cost frontier One model at each effort level
0.00 20.0 40.0 60.0 80.0 100.0 $10 $20 $50 $100 $200 $500 $1,000 Cost per task, log scale GPT-5.2 Codex, xhigh GPT-5.4, xhigh GPT-5.3 Codex GPT-5 mini, high Kimi K2.5 0.00 20.0 40.0 60.0 80.0 100.0 $10 $100 $1,000 Cost per task, log scale GPT-5.2 Codex, xhigh GPT-5.4, xhigh

Source: Official Letta Leaderboard leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.2 Codex xhigh 93.0 $44
GPT-5.4 xhigh 89.0 $44
Claude Sonnet 4.6 88.0 $116
GPT-5.2 xhigh 87.0 $77
GPT-5.3 Codex 85.0 $41
Claude Opus 4.6 84.0 $268
Gemini 3 Flash 82.0 $38
GPT-5 mini high 67.0 $36
Claude Opus 4.5 63.0 $568
Kimi K2.5 57.0 $27
Claude Sonnet 4.5 48.0 $405
Claude Haiku 4.5 43.0 $118
GLM-5 39.0 $86
MiniMax-M2.5 37.0 $50

The best score, 93.0, costs $44 a task

Runs on the cost frontier score higher than any run that's cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

OpenAI Google DeepMind Moonshot AI Cost frontier One model at each effort level
0.00 20.0 40.0 60.0 80.0 100.0 $10 $20 $50 $100 $200 $500 $1,000 Cost per task, log scale GPT-5.2 Codex, xhigh GPT-5.4, xhigh GPT-5.3 Codex GPT-5 mini, high Kimi K2.5 0.00 20.0 40.0 60.0 80.0 100.0 $10 $100 $1,000 Cost per task, log scale GPT-5.2 Codex, xhigh GPT-5.4, xhigh

Source: Official Letta Leaderboard leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-5.2 Codex xhigh 93.0 $44
GPT-5.4 xhigh 89.0 $44
Claude Sonnet 4.6 88.0 $116
GPT-5.2 xhigh 87.0 $77
GPT-5.3 Codex 85.0 $41
Claude Opus 4.6 84.0 $268
Gemini 3 Flash 82.0 $38
GPT-5 mini high 67.0 $36
Claude Opus 4.5 63.0 $568
Kimi K2.5 57.0 $27
Claude Sonnet 4.5 48.0 $405
Claude Haiku 4.5 43.0 $118
GLM-5 39.0 $86
MiniMax-M2.5 37.0 $50

3 models have held number one on the official leaderboard since its release in May 2025

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

OpenAI
Own leaderboard Not tested since Mar 2026

Source: Official Letta Leaderboard 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
20.0 40.0 60.0 80.0 100.0 2026 20.0 40.0 60.0 80.0 100.0 2026

Source: Official Letta Leaderboard 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.

More in Long context

All 115 in Long context
  • MRCR v2 In use Multi-round coreference: retrieve the n-th of several similar requests hidden in a long synthetic conversation, up to 1M tokens. Muse Spark 1.3 98.5%
  • LongBench v2 In use 503 multiple-choice questions over 8K-2M-word contexts (QA, dialogue, code repos, structured data). Accuracy. Inkling 2 Small 70.6%
  • GraphWalks In use A graph is given as an edge list in long context. The model answers BFS or parent queries, scored by F1. Claude Fable 5 99.96%
  • OpenAI MRCR v2 In use Models read a long synthetic chat that hides 2, 4 or 8 identical writing requests and must reproduce the response to the i-th one. Scored by mean string match ratio per context bin. Gemini 3.7 Flash 89%
  • CL-bench In use Model must learn new knowledge from a supplied long context (rules, procedures, data) and solve tasks. Strict LLM-verified rubrics. GPT-5.4 27.9%
  • AA-LCR 1.1 In use Models answer 100 questions that need reasoning across sets of real documents of about 100K tokens. An LLM equality checker with a fixed system prompt compares each answer with the key. Step 5 Preview 88.3%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 8 months ago, in March 2026, for GPT-5.4. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the official leaderboard (15 entries).
  • Pass Models scored 14 models we track, plus 1 other entry on its leaderboard.
  • Warning Newest score The newest came 8 months ago, in March 2026, for GPT-5.4.
  • Note Frontier models None of the top 10 has a score. The strongest tested is GPT-5.4 (#28).
  • Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
  • Pass Importer We read the board on docs.letta.com: 14 rows for models we track and 1 other entry at the last read.
  • Pass Last read November 21st, 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