GPT-5.4 vs GPT-OSS 120B

Updated October 17, 2026. Rank says GPT-5.4 is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.

Decision reading

GPT-5.4 has the higher public score, 66.7 versus 37.32, and the 90% score intervals do not overlap. 4 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Model A

GPT-5.4

OpenAI

66.7/100

Supported · Public rank #30

90% interval 62.2–71.2

Model B

GPT-OSS 120B

OpenAI

37.32/100

Supported · Public rank #148

90% interval 23.6–51.1

Shared results: 4
GPT-5.4 only: 38
GPT-OSS 120B only: 4
Like-for-like categories: 1 / 8

Supported: GPT-5.4 and GPT-OSS 120BHow the comparison works

Which one for your work

Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.

Long documents
Prompts that approach the documented context limit
GPT-5.4
GPT-5.4 has the larger documented context window.
Confidence: documented

Show secondary and unsupported calls

Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
GPT-OSS 120B is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
GPT-OSS 120B is scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.
Confidence: limited
Chat turn cost
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
The page does not recommend a cost winner because at least one model cannot fit the stated workload in one request. GPT-OSS 120B does not fit this workload in one request. GPT-OSS 120B has no comparable published API token rate.
Confidence: listed-rates
Repository review cost
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates

Which one for a specific job

Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.

Pick the job you would use them for

The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.

47.7GPT-5.423.0GPT-OSS 120B

Directional only · BenchAlign v5.8

GPT-5.4 has the higher coding point estimate. Conditional score ranges do not establish rank confidence.

Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.

Open the coding leaderboard Shortlist models for this job

Same basis rules as the category table below

Shape of the matched evidence

Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.

Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.

HLEKnowledge
Normalized gap 37.2
GPT-5.4:52.1%
GPT-OSS 120B:14.9%
GPT-5.4 source GPT-OSS 120B source
GPQAKnowledge
Normalized gap 12.7
GPT-5.4:92.8%
GPT-OSS 120B:80.1%
GPT-5.4 source GPT-OSS 120B source

Bars run 0–100 on each benchmark’s normalized display scale

Category results, on a stated basis

Each row shows the public-lane category score for both models: the BenchAlign v5.8 lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.

Knowledge

Like-for-like

GPT-5.4: 69.1; Supported · #20/184
GPT-OSS 120B: 38.6; Supported · #122/184
Basis: BenchAlign v5.8 lane · 7 vs 3 public rows
Reading: GPT-5.4 leads

Agentic

Directional only

GPT-5.4: 52.1; Supported · #44/128
GPT-OSS 120B: 19.5; Estimated · #110/128
Basis: BenchAlign v5.8 lane · 15 vs 1 public rows
Reading: Directional only

Coding

Directional only

GPT-5.4: 47.7; Supported · #48/150
GPT-OSS 120B: 23.0; Estimated · #111/150
Basis: BenchAlign v5.8 lane · 6 vs 2 public rows
Reading: Directional only

Instruction following

Directional only

GPT-5.4: 90.7; #23/148
GPT-OSS 120B: 84.3; #50/148
Basis: Provisional lane · 0 vs 0 weighted rows
Reading: Directional only

Reasoning

Not comparable

GPT-5.4: 64.2; #23/28
GPT-OSS 120B: 61.6; Unranked · 2 rankable rows
Basis: Provisional lane · 2 vs 0 weighted rows
Reading: Not comparable

Multimodal

Not comparable

GPT-5.4: 73.1; #30/54
GPT-OSS 120B: Not ranked
Basis: Provisional lane · 3 vs 0 weighted rows
Reading: Not comparable

Multilingual

Not comparable

GPT-5.4: Not ranked
GPT-OSS 120B: Not ranked
Basis: Provisional lane · 0 vs 0 weighted rows
Reading: Not comparable

Math

Not comparable

GPT-5.4: 82.0; Unranked · 2 rankable rows
GPT-OSS 120B: Not ranked
Basis: Provisional lane · 2 vs 0 weighted rows
Reading: Not comparable

Category scores with the server-provided evidence basis for GPT-5.4 and GPT-OSS 120B
Category	GPT-5.4	GPT-OSS 120B	Basis	Reading
Knowledge	69.1Supported · #20/184	38.6Supported · #122/184	Like-for-likeBenchAlign v5.8 lane · 7 vs 3 public rows	GPT-5.4 leads
Agentic	52.1Supported · #44/128	19.5Estimated · #110/128	Directional onlyBenchAlign v5.8 lane · 15 vs 1 public rows	Directional only
Coding	47.7Supported · #48/150	23.0Estimated · #111/150	Directional onlyBenchAlign v5.8 lane · 6 vs 2 public rows	Directional only
Instruction following	90.7#23/148	84.3#50/148	Directional onlyProvisional lane · 0 vs 0 weighted rows	Directional only
Reasoning	64.2#23/28	61.6Unranked · 2 rankable rows	Not comparableProvisional lane · 2 vs 0 weighted rows	Not comparable
Multimodal	73.1#30/54	Not ranked	Not comparableProvisional lane · 3 vs 0 weighted rows	Not comparable
Multilingual	Not ranked	Not ranked	Not comparableProvisional lane · 0 vs 0 weighted rows	Not comparable
Math	82.0Unranked · 2 rankable rows	Not ranked	Not comparableProvisional lane · 2 vs 0 weighted rows	Not comparable

Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign v5.8) differ from the provisional-lane categories. Unranked scores sit on the provisional lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.

Supported evidence per lane · bars run 0–100Methodology

What each workload costs

Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.

Chat turn

1K fresh input + 500 output tokens

GPT-5.4: $0.01; Fits in one request
GPT-OSS 120B: Self-hosted; infrastructure cost varies; Fits in one request

GPT-OSS 120B has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

GPT-5.4: $0.17; Fits in one request
GPT-OSS 120B: Self-hosted; infrastructure cost varies; Fits in one request

GPT-OSS 120B has no comparable published API token rate.

Cache-heavy agent loop

200K cached + 20K fresh input + 10K output tokens

GPT-5.4: $0.25; Fits in one request
GPT-OSS 120B: Self-hosted; infrastructure cost varies; Does not fit in one request; Cached-input rate unavailable

GPT-OSS 120B does not fit this workload in one request. GPT-OSS 120B has no comparable published API token rate.

Cached input falls back to the list input rate only where a cached rate is unpublished

Specification differences

Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.

SpecificationGPT-5.4GPT-OSS 120B

Context window

Maximum documented context; output-token limits may be lower.

GPT-5.4

1.05M

OpenAI pricing

GPT-OSS 120B

128K

API model ID

GPT-5.4

gpt-5.4

OpenAI pricing

GPT-OSS 120B

Not sourced

Cached-input rate

A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.

GPT-5.4

$0.25 per 1M cached input tokens

OpenAI pricing

GPT-OSS 120B

No comparable hosted API rate

Documented inputs

GPT-5.4

Not sourced

GPT-OSS 120B

Not sourced

Documented outputs

GPT-5.4

Not sourced

GPT-OSS 120B

Not sourced

Provider availability

GPT-5.4

Not sourced

GPT-OSS 120B

Not sourced

Reasoning profile

GPT-5.4

Reasoning

GPT-OSS 120B

Non-Reasoning

Weight access

GPT-5.4

Proprietary

GPT-OSS 120B

Open Weight

License

GPT-5.4

Proprietary

GPT-OSS 120B

Open Weight

Release date

GPT-5.4

2026-03-05

GPT-OSS 120B

2025-08-05

If you already use one of these models

Deployment change: Both entries list OpenAI as the provider. Confirm endpoint, model ID, limits, and feature support before switching.
Quality signal: GPT-5.4 has the higher public score, 66.7 versus 37.32, and the 90% score intervals do not overlap.
Workload cost: A complete comparable API-rate estimate is not available for both models.
Context tradeoff: GPT-5.4 has the larger documented window (1.05M).

Run the same representative tasks against both endpoints before changing production traffic.

Questions

Which is better, GPT-5.4 or GPT-OSS 120B?

GPT-5.4 has the higher public score, 66.7 versus 37.32, and the 90% score intervals do not overlap. That is the clearest overall quality signal in the current public data.

Which is better for coding, GPT-5.4 or GPT-OSS 120B?

GPT-5.4 scores higher for coding on the public lane, 47.7 to 23. GPT-OSS 120B is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like. BenchLM does not name a winner for coding from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which is better for agentic tasks, GPT-5.4 or GPT-OSS 120B?

GPT-5.4 scores higher for agentic tasks on the public lane, 52.1 to 19.5. GPT-OSS 120B is scored on Estimated evidence for agentic tasks, so the reading is directional rather than like-for-like. BenchLM does not name a winner for agentic tasks from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which costs less, GPT-5.4 or GPT-OSS 120B?

Both models do not have comparable published API token rates, so this page does not name a universal price winner.

Which has the larger context window, GPT-5.4 or GPT-OSS 120B?

GPT-5.4 has the larger documented context window: 1.05M, compared with 128K.

Self-host vs API cost

Estimates at 50,000 req/day · 1000 tokens/req average.

GPT-5.4

API / mo$13,125

Self-host / moNot listed

Break-even—

Proprietary model — self-hosting not applicable.

GPT-OSS 120B

API / mo$0

Self-host / mo$3,495

Break-even—

Model the full break-even

Benchmark evidence

The full public result ledger is available for audit without forcing a wide desktop table onto a phone.

Browse raw public benchmark evidence46 rows

Agentic

Terminal-Bench 2.0
GPT-5.475.1%
Source
GPT-OSS 120B—
Not directly comparable
CyberGym
GPT-5.479.0%
Source
GPT-OSS 120B—
Not directly comparable
BrowseComp
GPT-5.482.7%
Source
GPT-OSS 120B—
Not directly comparable
OSWorld-Verified
GPT-5.475%
Source
GPT-OSS 120B—
Not directly comparable
MCP Atlas
GPT-5.470.6%
Source
GPT-OSS 120B—
Not directly comparable
Toolathlon
GPT-5.454.6%
Source
GPT-OSS 120B—
Not directly comparable
τ²-bench results
GPT-5.498.9%
Source
GPT-OSS 120B—
Not directly comparable
Claw-Eval
GPT-5.460.3%
Source
GPT-OSS 120B—
Not directly comparable
DeepSearchQA
GPT-5.473.6%
Source
GPT-OSS 120B—
Not directly comparable
Gert Labs
Shared source
GPT-5.464.89%
GPT-OSS 120B29.61%
GPT-5.4 leads this result
ResearchClawBench
GPT-5.415.3%
Source
GPT-OSS 120B—
Not directly comparable
JobBench
GPT-5.438.9%
Source
GPT-OSS 120B—
Not directly comparable
ExploitGym
GPT-5.46.0%
Source
GPT-OSS 120B—
Not directly comparable
ApprenticeBench
GPT-5.411%
Source
GPT-OSS 120B—
Not directly comparable
MazeBench
GPT-5.40%
Source
GPT-OSS 120B—
Not directly comparable

Coding

LiveCodeBench Pro
GPT-5.487.5%
Source
GPT-OSS 120B—
Not directly comparable
SWE-bench Pro
GPT-5.457.7%
Source
GPT-OSS 120B—
Not directly comparable
React Native Evals
Shared source
GPT-5.485.3%
GPT-OSS 120B71.6%
GPT-5.4 leads this result
Terminal-Bench 2.0
GPT-5.475.1%
Source
GPT-OSS 120B—
Not directly comparable
Vibe Code Bench
GPT-5.467.42%
Source
GPT-OSS 120B—
Not directly comparable
PostTrainBench v1.1
GPT-5.419.0%
Source
GPT-OSS 120B—
Not directly comparable
SWE-bench Verified
GPT-5.4—
GPT-OSS 120B62.4%
Source
Not directly comparable

Reasoning

ARC-AGI-2
GPT-5.474.0%
Source
GPT-OSS 120B—
Not directly comparable
ARC-AGI-3
GPT-5.40.2%
Source
GPT-OSS 120B—
Not directly comparable
ARC-AGI-1
GPT-5.493.67%
Source
GPT-OSS 120B—
Not directly comparable

Multimodal

MMMU-Pro
GPT-5.481.2%
Source
GPT-OSS 120B—
Not directly comparable
OfficeQA Pro
GPT-5.453.2%
Source
GPT-OSS 120B—
Not directly comparable
MMMU-Pro w/ Python
GPT-5.482.1%
Source
GPT-OSS 120B—
Not directly comparable
CharXiv
GPT-5.482.8%
Source
GPT-OSS 120B—
Not directly comparable
ERQA
GPT-5.465.4%
Source
GPT-OSS 120B—
Not directly comparable
SimpleVQA
GPT-5.461.1%
Source
GPT-OSS 120B—
Not directly comparable
ScreenSpot Pro
GPT-5.485.4%
Source
GPT-OSS 120B—
Not directly comparable
ZeroBench
GPT-5.441.0%
Source
GPT-OSS 120B—
Not directly comparable
MedXpertQA (MM)
GPT-5.477.1%
Source
GPT-OSS 120B—
Not directly comparable

Knowledge

GPQA
GPT-5.492.8%
Source
GPT-OSS 120B80.1%
Source
GPT-5.4 leads this result
HLE
GPT-5.452.1%
Source
GPT-OSS 120B14.9%
Source
GPT-5.4 leads this result
HLE w/o tools
GPT-5.439.8%
Source
GPT-OSS 120B—
Not directly comparable
GPQA-D
GPT-5.492.8%
Source
GPT-OSS 120B—
Not directly comparable
HealthBench Hard
GPT-5.440.1%
Source
GPT-OSS 120B—
Not directly comparable
MedXpertQA (Text)
GPT-5.459.6%
Source
GPT-OSS 120B—
Not directly comparable
HealthBench Professional
GPT-5.448.1%
Source
GPT-OSS 120B—
Not directly comparable
MMLU
GPT-5.4—
GPT-OSS 120B90%
Source
Not directly comparable

Math

FrontierMath v2 (Tiers 1-3)
GPT-5.478.596%
Source
GPT-OSS 120B—
Not directly comparable
FrontierMath v2 (Tier 4)
GPT-5.449.000%
Source
GPT-OSS 120B—
Not directly comparable
AIME 2024
GPT-5.4—
GPT-OSS 120B95.8%
Source
Not directly comparable
AIME 2025
GPT-5.4—
GPT-OSS 120B92.5%
Source
Not directly comparable

46 public results · 4 shared

Related comparisons

Compare API pricing Read the methodology Open the model selector

Watch GPT-5.4 vs GPT-OSS 120B

One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.

Read a sample issue

Join 5,600+ readers.

Last updated October 17, 2026

GPT-5.4 vs GPT-OSS 120B

Decision reading

Which one for your work

Long documents

Coding work

Agentic work

Chat turn cost

Cache-heavy agent loop cost

Repository review cost

Which one for a specific job

What is actually comparable

Shape of the matched evidence

Category results, on a stated basis

Knowledge

Agentic

Coding

Instruction following

Reasoning

Multimodal

Multilingual

Math

What each workload costs

Chat turn

Repository review

Cache-heavy agent loop

Specification differences

Context window

API model ID

Cached-input rate

Documented inputs

Documented outputs

Provider availability

Reasoning profile

Weight access

License

Release date

If you already use one of these models

Questions

Which is better, GPT-5.4 or GPT-OSS 120B?

Which is better for coding, GPT-5.4 or GPT-OSS 120B?

Which is better for agentic tasks, GPT-5.4 or GPT-OSS 120B?

Which costs less, GPT-5.4 or GPT-OSS 120B?

Which has the larger context window, GPT-5.4 or GPT-OSS 120B?

Self-host vs API cost

Benchmark evidence

Agentic

Coding

Reasoning

Multimodal

Knowledge

Math

Related comparisons

Watch GPT-5.4 vs GPT-OSS 120B