AgentBench / results
← All models

GPT-OSS 120B (stock)

1 tested weight format · Explore results, then inspect the answers.

Choose a tested configuration

Get this tested model

Q4_K_M · 62.77 GB download · 2 files required

Links follow the selected quant and configuration.

Setup, memory & provenance · Q4_K_M

Q4_K_M · 62.77 GB of weights · rocm · context 16384 · GPU layers 99

Large · 20B and above · MoE · 116.8B total · Active parameter count not recorded

Memory fit has not been verified for this configuration.

completed

Recorded settings and thinking verification

Temperature 0 · seed 42 · output limit 4096 · reasoning effort medium

RAM labels describe the test computer, not this model’s memory requirement. Speed comparisons keep the 32 GB and 128 GB runs separate.

Runtime build: unrecorded · Local · test machine with 128 GB RAM · concurrency: 1

Thinking status: native reasoning · probe 3 · scope Profile-bound diagnostic confirms native reasoning; separate checkpoint, no verified off condition..

Requested off does not prove an absence of reasoning. Saved checks do not establish unrecorded historical template or build details.

Overall across tasks

73.8 / 100

Nine categories, equally weighted · 9/9 eligible

Complete coverage for this setup. Quality only; speed is separate. How the index works →

What does it do well?

Select a category to explore the answers. Bars show what happened on every planned task. Incorrect answers and failures to finish both earn zero; technical problems and untested tasks stay unscored.

Exact scores and task times by suite

Each suite has its own metric and fixed task bank. The balanced index weights categories equally; individual scores and evidence remain available below.

Test suiteReasoning · mediumDetails
Everyday coding tasksTask pass rate
Pass rate
14/16 · 88%
Answers completed
16/16 · 100%
Correct / completed
14/16 · 88%
14 full credit2 incorrect
16/16 scored49.4s median task time · 16 measured
Larger practical tasksMean task credit
Average credit
54.5%
Full-credit rate
6/11 · 55%
Answers completed
11/11 · 100%
Full credit / completed
6/11 · 55%
6 full credit5 incorrect
11/11 scored99.4s median task time · 11 measured
Logic and number puzzlesMean puzzle credit
Average credit
69.8%
Full-credit rate
23/34 · 68%
Answers completed
34/34 · 100%
Full credit / completed
23/34 · 68%
23 full credit1 partial credit10 incorrect
34/34 scored153.2s median task time · 34 measured
Published hard problemsMean answer credit
Average credit
46.7%
Full-credit rate
5/11 · 45%
Answers completed
11/11 · 100%
Full credit / completed
5/11 · 45%
5 full credit1 partial credit5 incorrect
11/11 scored170.1s median task time · 11 measured
Following exact instructions · original campaignStrict prompt complianceNot planned
Reading without guessingExact-match accuracy
Pass rate
13/16 · 81%
Answers completed
16/16 · 100%
Correct / completed
13/16 · 81%
13 full credit3 incorrect
16/16 scored11.1s median task time · 16 measured
Calculating from a passageNumeric exact-match accuracy
Pass rate
7/7 · 100%
Answers completed
7/7 · 100%
Correct / completed
7/7 · 100%
7 full credit
7/7 scored14.7s median task time · 7 measured
Reading research papersAnswer token F1
Average credit
33.5%
Full-credit rate
0/8 · 0%
Answers completed
8/8 · 100%
Full credit / completed
0/8 · 0%
8 partial credit
8/8 scored32.0s median task time · 8 measured
Resisting incorrect correctionsAll-branches pass rate
Pass rate
12/12 · 100%
Answers completed
12/12 · 100%
Correct / completed
12/12 · 100%
12 full credit
12/12 scored31.0s median task time · 12 measured
Following exact instructions · 128 GB campaignStrict prompt compliance
Pass rate
21/23 · 91%
Answers completed
23/23 · 100%
Correct / completed
21/23 · 91%
21 full credit2 incorrect
23/23 scored64.7s median task time · 23 measured

Model analysis

This analysis covers this configuration across its recorded thinking conditions and test suites.

Use this stock Q4_K_M configuration for small project edits and source-based assistance, rather than unattended research or larger projects. With native reasoning at medium effort, it handled routine coding well and resisted incorrect corrections while accepting genuine source updates. Larger practical work was uneven, and research-paper answers were weak. Small coding tasks took a median 49.4 seconds on this test machine.

Evidence for this judgment
Everyday coding tasks · Reasoning · medium
Tasks passed14 of 16 (87.5%)
Test coverage16 of 16 tasks scored
Tasks timed16 tasks
Typical time per task (median)49.4 seconds
Explore these test results →

View this model’s answers, code and tool activity →

Resisting incorrect corrections · Reasoning · medium
Tasks passed12 of 12 (100.0%)
Test coverage12 of 12 tasks scored
Explore these test results →

View this model’s answers, code and tool activity →

Larger practical tasks · Reasoning · medium
Average score54.5/100 · Mean task credit
Test coverage11 of 11 tasks scored
Task outcomes6 full credit · 0 partial · 5 no credit
Explore these test results →

View this model’s answers, code and tool activity →

Reading research papers · Reasoning · medium
Average score33.5/100 · Answer token F1
Test coverage8 of 8 tasks scored
Task outcomes0 full credit · 8 partial · 0 no credit
Explore these test results →

View this model’s answers, code and tool activity →

Summary generated by GPT Astra · · Edited by Codex; agent reviewed by Codex root · publication authorized by user

About this summary
Analyst provider
openai
Generation route
Codex CLI · codex-cli 0.163.0 · ChatGPT account
Requested model
gpt-6-astra
Returned model
Not supplied; identity is recorded from the request only.
Exact provider snapshot
Exact provider snapshot not supplied.
Reasoning effort
high
Prompt version
model-summary-reader-v8
Generation time
2026-10-28T01:06:30.149176Z
Review time
2026-10-28T01:07:17.281030+00:00
Evidence reviewer
Codex root (agent)
Publication authorization
User · 2026-10-22T05:55:43.625712+00:00 · Go ahead and complete the remaining work. Spin up agents: luna, terra, sol, astra, as needed to get it done. Narrative summaries: these are the astra generated summaries? Go ahead and get those completed as well. Once all of this is done, you may push the completed result.
Originating evidence edition
dataset-161a0c0b3366764cd46aca8c85ef893157074bade44c7f42829d071766da607e
Analysis input hash
b4ee3d7533a337fe8dbd80cef9b41320ad3c405abb158a0d49dbdbde1d02f122
Summary revision
summary-bd1ae69c75882ac878822c791fb2f364d28b303733ed3ffac91e637a27c1e2de

Generation and editorial authorship remain attached to this revision. Changing the default generator does not change this attribution.

Good at

Handles small project edits and tightly constrained responses well. Strong results on routine coding and wording, language and formatting requirements make it useful for bounded assistance, with checks still advisable.

Evidence for this judgment
Everyday coding tasks · Reasoning · medium
Tasks passed14 of 16 (87.5%)
Test coverage16 of 16 tasks scored
Explore these test results →

View this model’s answers, code and tool activity →

Following exact instructions · Reasoning · medium
Tasks passed21 of 23 (91.3%)
Test coverage23 of 23 tasks scored
Explore these test results →

View this model’s answers, code and tool activity →

Works well with short source passages: answering questions, calculating from their facts, and resisting incorrect corrections while accepting genuine source updates. These small tests support passage-based assistance, not a blanket endorsement for research.

Evidence for this judgment
Resisting incorrect corrections · Reasoning · medium
Tasks passed12 of 12 (100.0%)
Test coverage12 of 12 tasks scored
Explore these test results →

View this model’s answers, code and tool activity →

Calculating from a passage · Reasoning · medium
Tasks passed7 of 7 (100.0%)
Test coverage7 of 7 tasks scored
Explore these test results →

View this model’s answers, code and tool activity →

Reading without guessing · Reasoning · medium
Tasks passed13 of 16 (81.3%)
Test coverage16 of 16 tasks scored
Explore these test results →

View this model’s answers, code and tool activity →

Struggles with

Research-paper answers need substantial verification. Their low answer matching score reflects weak answer quality despite finishing the tasks; success with short passages did not carry over to this more demanding reading.

Evidence for this judgment
Reading research papers · Reasoning · medium
Average score33.5/100 · Answer token F1
Test coverage8 of 8 tasks scored
Task outcomes0 full credit · 8 partial · 0 no credit
Partially correct answers8 tasks
Failed to finish0 tasks
Explore these test results →

View this model’s answers, code and tool activity →

Larger practical tasks and difficult math/reasoning need close checking: many finished outputs were incorrect, rather than unfinished. Logic and number puzzles were also uneven, so native reasoning is no guarantee of correctness.

Evidence for this judgment
Larger practical tasks · Reasoning · medium
Average score54.5/100 · Mean task credit
Test coverage11 of 11 tasks scored
Task outcomes6 full credit · 0 partial · 5 no credit
Incorrect completed answers5 tasks
Failed to finish0 tasks
Explore these test results →

View this model’s answers, code and tool activity →

Published hard problems · Reasoning · medium
Average score46.7/100 · Mean answer credit
Test coverage11 of 11 tasks scored
Task outcomes5 full credit · 1 partial · 5 no credit
Incorrect completed answers5 tasks
Failed to finish0 tasks
Explore these test results →

View this model’s answers, code and tool activity →

Logic and number puzzles · Reasoning · medium
Average score69.8/100 · Mean puzzle credit
Test coverage34 of 34 tasks scored
Task outcomes23 full credit · 1 partial · 10 no credit
Explore these test results →

View this model’s answers, code and tool activity →

Worth knowing

This stock configuration uses native reasoning at medium effort. Without a matched thinking-off test, its results cannot establish whether reasoning improves quality enough to justify its time cost.

Evidence for this judgment

Source limitation
This is a native reasoning condition at its recorded effort, not a matched thinking-on/off experiment.

Source limitation
Stock, non-abliterated GPT-OSS 120B uses native reasoning at medium effort. No matched thinking-off condition exists in this campaign.

Publication redactions limit reader inspection of some evidence.

Evidence for this judgment

Source limitation
Some selected evidence has recorded publication redactions.

How to interpret these assessments and their limitations →

One weight format is recorded for this model. Additional tested formats will appear here when available.

What changed in these results

Stock, non-abliterated GPT-OSS 120B uses native reasoning at medium effort. No matched thinking-off condition exists in this campaign.

Thinking tradeoff

No verified on/off comparison is available for this configuration and test suite. Requested modes alone do not establish a controlled thinking comparison.

Sampled repetitions

Repeated runs are a separate experiment and never add votes to the baseline score.

Reasoning · medium · temperature 0.6

48/48 repeats scored · 41 full-credit repeats.

16/16 questions have every planned repeat scored. Of those, 13 earned full credit on every repeat.

Recorded repetition conditions
  • Seed 202 · requested on · native reasoning
  • Seed 101 · requested on · native reasoning
  • Seed 303 · requested on · native reasoning

Everyday coding tasks: recorded outcomes

Find information in a small project, edit files and fix bugs.

14 full credit2 incorrect

Each segment counts tasks, not credit points. Partial credit varies by task.

Scoring and fixed question bank →

This model’s answers, code and tool activity

Select a question to read the original answer, tool calls and available code changes for this configuration and thinking setting.

1–16 of 16 matching questions · 16 in this version

  • Locate the Retry Limit

    The user asks where the project's retry-limit setting is defined and what value it holds. This exercises locating a named setting in code and reporting both its location and value.

    basic-T1 · primary
  • Find the Release Date

    The user asks for the release date mentioned in a project notes document. This exercises reading a specified file and extracting a particular fact rather than guessing from outside knowledge.

    basic-T2 · primary
  • Change the Retry Limit

    The user asks to change the project's retry limit to a specified value. This exercises finding the relevant setting and making a focused code edit rather than merely describing the change.

    basic-T3 · primary
  • Create an Exact One-Line File

    The user asks to create a text file containing exactly one supplied line. This exercises basic file creation and precise instruction following, including avoiding extra text in the file.

    basic-T4 · primary
  • Explain the Latest Commit

    The user asks what the most recent Git commit changed in the project. This exercises inspecting recorded changes and turning them into a clear explanation of what was modified.

    basic-T5 · primary
  • Count and Compare Python Files

    The user asks how many Python files are in the application folder and which is largest. This exercises identifying relevant files, counting them, and comparing their sizes.

    basic-T6 · primary
  • Add Readable Byte Sizes

    The user asks to add a function that converts byte counts into readable sizes and remove the related TODO comment. This exercises implementing a small utility and updating the surrounding code consistently.

    basic-T7 · primary
  • Collect TODO Comments

    The user asks to find every TODO comment in the Python code and append each to the project's task list on a separate line. This combines searching across files with collecting text into an existing document.

    basic-T8 · primary
  • Fix Code Behind a Failing Test

    The user asks to investigate a failing test and fix the application code without changing the test. This exercises diagnosing a bug while respecting an explicit limit on what may be edited.

    basic-H1 · primary
  • Rename a Function Throughout

    The user asks to rename a network function and update every place in the project that uses it. This exercises tracking references across files and making a coordinated change rather than editing only the definition.

    basic-H2 · primary
  • Find the Proxy Settings

    The user asks to find a proxy configuration file somewhere under the project's parent folder and report its host and port. This exercises searching beyond the project folder and reading configuration values.

    basic-H3 · primary
  • Locate an Error in a Log

    The user asks which error code appears in a long event log and on which line. This exercises finding a specific entry in a large file and reporting its precise location.

    basic-H4 · primary
  • Append a Final Task

    The user asks to append a supplied release-related task as the final line of an existing task list. This exercises preserving existing content while adding exact text in the requested position.

    basic-H5 · primary
  • Calculate Total Retry Waiting

    The user asks for the total time a network function spends sleeping when all attempts fail, worked out from the code. This exercises following retry logic and adding the delays along that failure path.

    basic-H6 · primary
  • Document Every Command-Line Option

    The user asks for a project readme listing every command-line option with its help text. This exercises discovering the program's available options and transferring that information into complete, source-grounded documentation.

    basic-H7 · primary
  • Undo a Commit by Editing

    The user asks to undo the latest Git commit's change by editing the file directly, without using Git to make changes. This exercises understanding a previous edit and reversing it under a method constraint.

    basic-H8 · primary
Filters change this list, not the fixed scoring denominator. Search currently covers question IDs, titles and available task explanations. Focus a control to load prompt/outcome search where this edition provides it.

Suggested comparisons

These are starting points for inspection, not claims that one setup is better. Runtime and placement differences remain visible.

No matching suggestion is available for this selection.

Tested weights

Q4_K_M · 62.77 GB

Publisher unresolved · legacy partial

gpt-oss-120b-Q4_K_M-00001-of-00002.ggufHistorical first-MiB hash only; not a full-file identity.

gpt-oss-120b-Q4_K_M-00002-of-00002.ggufHistorical first-MiB hash only; not a full-file identity.

Explore the answers

Choose a suite to inspect its exact available questions, references and model answers.