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 32 GB RAM · concurrency: 1
Thinking status: verified · probe 1 · scope Saved profile-level probe; exact historical template/build scope is not fully recorded..
Requested off does not prove an absence of reasoning. Saved checks do not establish unrecorded historical template or build details.
Overall across tasks
77.9 / 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.
This analysis covers this configuration across its recorded thinking conditions and test suites.
Use this model for small project edits with thinking off, or longer practical tasks with thinking on. It handled all the assessed small coding tests and most larger tasks in those modes, so it is more than a simple coding assistant. Difficult math and reasoning were weak, however, and strict formatting still needs checking.
Original analysis generated by GPT Astra · · Edited by Codex; agent reviewed by Codex · publication authorized by user
About this summary
Codex reviewed this assessment against a newer evidence edition. The original analyst, generation date and source draft remain recorded separately from the editorial revision.
Analyst provider
openai
Generation route
Codex CLI · codex-cli 0.162.1 · 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-v6
Generation time
2026-10-12T05:53:56.011619Z
Review time
2026-10-22T07:26:27.217016+00:00
Evidence reviewer
Codex (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.
Handles small project edits and bug fixes well: all small coding tasks passed with thinking off. That supports everyday assistance, not a promise that arbitrary projects will work.
Can carry out longer tasks with tools: with thinking on, most larger practical tasks passed completely. It also passed every source-backed challenge test, retaining correct answers while accepting genuine source updates.
Difficult math and reasoning are a poor fit: average answer scores were low with thinking both on and off. Successful coding results should not be taken as evidence of strong advanced problem-solving.
Can get stuck repeating tool calls rather than changing approach. In the examples checked, duplicate-call warnings did not always stop repetition—even during a larger task that ultimately succeeded.
Evidence for this judgment
Example 1: Larger practical tasks Thinking off. Final result: no credit. A final score does not describe every tool step; this example supports a behavior observation, not a whole-test conclusion. View the question and recorded answer →
Example 2: Larger practical tasks Thinking on. Final result: full credit. A final score does not describe every tool step; this example supports a behavior observation, not a whole-test conclusion. View the question and recorded answer →
Source limitation Qualitative examples are a deterministic sample; they do not establish whole-bank behavior.
Should you enable thinking?
Leave thinking off for everyday coding: paired tests showed no accuracy gain but longer task times. It also reduced puzzle scores while taking longer, so extra deliberation is not a general upgrade.
Enable thinking for longer tool-driven work and for keeping source-backed answers correct under challenge. Both improved substantially in paired tests, making the longer task times worthwhile for those uses.
Evidence for this judgment
Larger practical tasks · Thinking off versus on
Thinking off → on
45.5/100 · Mean task credit → 90.9/100 · Mean task credit
With thinking off, small coding produced the best pass result in this three-configuration comparison, though both alternatives finished sooner; Qwen3.5 4B was closer in score, while Gemma 4 E4B scored much lower.
Evidence for this judgment
Speed: Everyday coding tasks Thinking off. Matching test-machine label and execution settings; spans model sizes. Timings include unsuccessful tasks, so read them alongside quality.
Speed and task success: Everyday coding tasksThinking off · Matching test-machine label and settings · Across model sizes
Shorter task time is better only when the result is useful. These times include failures, tools and grading. Token generation speed measures a different part of the work.
Results from this test setup, not a promise for your computer. Hardware, software, tokenizers and answer length can change the ordering. Historical runtime builds were not always captured. Comparison saved with the 2026-10-12 assessment.
Test machine: AMD Ryzen 5 5600X (6 cores / 12 threads) · AMD Radeon RX 6700 XT (12 GB VRAM) · 32 GB or 128 GB system RAM, never mixed in one timing group. Full test machine details
With thinking off, its puzzle credit led both matched alternatives. Task time fell between them: slower than the closer-scoring Qwen3.5 4B and faster than the lower-scoring Gemma 4 E4B.
Evidence for this judgment
Speed: Logic and number puzzles Thinking off. Matching test-machine label and execution settings; spans model sizes. Timings include unsuccessful tasks, so read them alongside quality.
Speed and task success: Logic and number puzzlesThinking off · Matching test-machine label and settings · Across model sizes
Shorter task time is better only when the result is useful. These times include failures, tools and grading. Token generation speed measures a different part of the work.
Results from this test setup, not a promise for your computer. Hardware, software, tokenizers and answer length can change the ordering. Historical runtime builds were not always captured. Comparison saved with the 2026-10-12 assessment.
Test machine: AMD Ryzen 5 5600X (6 cores / 12 threads) · AMD Radeon RX 6700 XT (12 GB VRAM) · 32 GB or 128 GB system RAM, never mixed in one timing group. Full test machine details
Worth knowing
Publication redactions, withheld artifacts and source-link-only paper exports limit readers’ inspection of some examples; they do not establish that the model lacked those inputs.
Evidence for this judgment
Source limitation Some selected evidence has recorded publication redactions.
One same-budget retry after context exhaustion. A second context limit is failed to answer (zero task-completion credit), not an incorrect submitted answer. Answer-only averages exclude these no-answer outcomes; completed retry answers enter the normal scoring path. This follow-up is selected after observing failures.
Evidence for this judgment
Source limitation One same-budget retry after context exhaustion. A second context limit is failed to answer (zero task-completion credit), not an incorrect submitted answer. Answer-only averages exclude these no-answer outcomes; completed retry answers enter the normal scoring path. This follow-up is selected after observing failures.
Everyday coding tasks · Task pass rate · requested / displayed reasoning on. Size is the weight download, not RAM. Different runtimes and placement can also affect results.
UD-Q2_K_XL · 4.12 GBLocal · test machine with 32 GB RAM · rocm · 99 GPU layers · Recorded · some results unscored or qualifiedThinking on14/16 passed16/16 scored
Qualified rows show the reason they are outside the rankings; they cannot establish superiority over a fully tested format. Open the setup to inspect context, runtime and controls.
Choosing a quant
Shared assessment of Q6_K, UD-Q2_K_XL, UD-Q3_K_XL, UD-Q4_K_XL, UD-Q5_K_XL, UD-Q6_K_XL · 2026-10-22 UTC. The current results table can include newer formats.
The selected weight format was not included in this written comparison. Inspect its recorded results separately.
UD-Q4_K_XL is a sensible starting point for small coding tasks. With reasoning requested off, it earned full credit alongside Q6_K, UD-Q5_K_XL and UD-Q6_K_XL, with a smaller download. UD-Q3_K_XL has a smaller download still and finished close behind requested-off; requested-on, it earned full credit alongside UD-Q4_K_XL and Q6_K. These narrow differences provide little reason to favor the largest download for this task bank.
Evidence for this judgment
Everyday coding tasks · Thinking requested off
Matching task controls and 16,384 context. These quality results describe the recorded setups; they do not isolate the effect of bit depth.
Weight format
Task pass rate
Task time
Download
Inspect
Q6_Kunslothrocm · 32 GB test machineCPU MoE layers: Not recordedRecorded execution settings
For requested-off larger tool-driven tasks, UD-Q4_K_XL is a stronger choice than UD-Q3_K_XL: its task-credit lead was substantial here. Its lead over Q6_K was only a single fully credited task, but its download is smaller.
Evidence for this judgment
Larger practical tasks · Thinking requested off
Matching task controls and 16,384 context. These quality results describe the recorded setups; they do not isolate the effect of bit depth.
Weight format
Mean task credit
Task time
Download
Inspect
Q6_Kunslothrocm · 32 GB test machineCPU MoE layers: Not recordedRecorded execution settings
For requested-on larger tasks, UD-Q6_K_XL led UD-Q3_K_XL and Q6_K by only a single fully credited task. UD-Q3_K_XL is a useful download-size tradeoff: it tied Q6_K on measured task credit with a smaller download and shorter median task time.
Evidence for this judgment
Larger practical tasks · Thinking requested on
Matching task controls and 16,384 context. These quality results describe the recorded setups; they do not isolate the effect of bit depth.
Weight format
Mean task credit
Task time
Download
Inspect
Q6_Kunslothrocm · 32 GB test machineCPU MoE layers: Not recordedRecorded execution settings
For requested-on research-paper questions, consider UD-Q5_K_XL: it earned higher answer credit than Q6_K and UD-Q6_K_XL while requiring a smaller download. This supports a task-specific choice, not a universal ranking.
Evidence for this judgment
Reading research papers · Thinking requested on
Matching task controls and 16,384 context. These quality results describe the recorded setups; they do not isolate the effect of bit depth.
Weight format
Answer token F1
Task time
Download
Inspect
Q6_Kunslothrocm · 32 GB test machineCPU MoE layers: Not recordedRecorded execution settings
These are comparisons of tested setups registered to one checkpoint. They do not isolate bit depth from conversion, publisher, runtime-build or template differences.
Only complete eligible banks with matching task controls and recorded context enter the quality comparison. Execution profiles are shown per setup; timing is compared only when every recorded execution profile matches and timing coverage is complete. Other results remain coverage, not ranked evidence.
Download size is not RAM or VRAM requirement. No memory-fit recommendation is established by this comparison.
Small score differences can reflect a single question. Similar results are not proof of equivalence; timings describe this benchmark machine.
Quant comparison generated by GPT Astra · 2026-10-22 UTC · Edited by Codex; agent reviewed by Codex · publication authorized by user
About this quant assessment
Requested model
gpt-6-astra
Returned model
Not supplied; request identity only.
Exact provider snapshot
Not supplied.
Generation route
Codex CLI · ChatGPT account
Prompt version
quant-summary-v3
Review time
2026-10-22T06:58:37.216745+00:00
Evidence reviewer
Codex (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.
One same-budget retry after context exhaustion. A second context limit is failed to answer (zero task-completion credit), not an incorrect submitted answer. Answer-only averages exclude these no-answer outcomes; completed retry answers enter the normal scoring path. This follow-up is selected after observing failures.
Both modes earned full credit on every matched question. These matched results leave no measured quality headroom; unmatched questions are not covered by this observation.
Thinking offMean matched task credit100.0%
Thinking onMean matched task credit100.0%
Thinking on took 7.3 seconds more per matched timed question, on average.
All matched quality and timing measurements
Matched measurement
Recorded result
Average credit, off → on
100.0% → 100.0%
Mean paired credit change
0.0 percentage points
Question outcomes
0 improved · 16 unchanged · 0 worse
Mean added task time
+7.3s
Mean task time, off → on
7.1s → 14.4s
Median task time, off → on
7.1s → 12.3s
Median of paired time changes
+5.1s
Total added task time
+117.3s
Ratio of summed times, on / off
2.0×
Quality and timing can cover different subsets. Median paired change is calculated per question; it is not the difference between the two medians. Task time includes tools and grading.
All tested weight formats of this checkpoint, using the selected thinking setting. This shared comparison updates with new results; each saved model summary keeps its original date.
Everyday coding tasks
Scores and times describe this task bank. Quick failures are not a speed advantage. Different execution settings are labeled.
Version
Task result
Typical task time
Weight download
Explore
Qwen3.5 9B Q6_KThinking on · Different or uncertain execution settings
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
Thinking on versus off
Verified requested on/off conditions of the same configuration, matched on the recorded test protocol.
Requested thinking off · Complete compared coverage