ZMIME
Comparison · 3 models · Updated Oct 4, 2026

GPT OSS 120B vs GPT OSS 20B vs Qwen3.5 9B

Qwen3.5 9B comes out ahead, 72 to 64 and 61 on our weighted score, and it is the cheaper option too.

  1. OpenAI

    GPT OSS 120B

    Released Aug 5, 2025

    61/100
    • ECI140.0
    • Price$0.15 / $0.60
    • Context131K
    Full specs Remove GPT OSS 120B

  2. OpenAI

    GPT OSS 20B

    Released Aug 5, 2025

    64/100
    • ECI137.8
    • Price$0.07 / $0.295
    • Context131K
    Full specs Remove GPT OSS 20B
  3. Our pick

    Alibaba (Qwen)

    Qwen3.5 9B

    Released Feb 23, 2026

    72/100
    • ECI139.5
    • Price$0.10 / $0.15
    • Context262K
    Full specs Remove Qwen3.5 9B
01 — Verdict

Qwen3.5 9B is our pick

Qwen3.5 9B is the better all-round choice, scoring 72/100 against GPT OSS 20B (64) and GPT OSS 120B (61). It leads on price, inputs & features and context window. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.

  • CapabilityGPT OSS 120BCapabilities Index (ECI): GPT OSS 120B 140.0 · Qwen3.5 9B 139.5 · GPT OSS 20B 137.8
  • Lowest priceQwen3.5 9BQwen3.5 9B $0.113 · GPT OSS 20B $0.126 · GPT OSS 120B $0.263 per 1M tokens (3:1 blend)
  • Longest contextQwen3.5 9BQwen3.5 9B 262,144 · GPT OSS 120B 131,072 · GPT OSS 20B 131,072 tokens
  • Widest inputsQwen3.5 9BGPT OSS 120B: Text · GPT OSS 20B: Text · Qwen3.5 9B: Text, Images, Video
  • Self-hostingAll open weightsEvery model here can be downloaded and run on your own hardware
How the score is built
MeasureWeightGPT OSS 120BGPT OSS 20BQwen3.5 9B
CapabilityCapabilities Index (ECI)50%656365
Price25%779295
Inputs & features15%454580
Context window10%242437
Overall100%61/10064/10072/100
02 — Side by side

Every spec in one table

Highlighted cells lead their row. Dashes mean the data is not published.

GPT OSS 120B vs GPT OSS 20B vs Qwen3.5 9B specifications side by side
SpecificationGPT OSS 120BOpenAIGPT OSS 20BOpenAIQwen3.5 9BAlibaba (Qwen)
Capability
Capabilities Index (ECI)140.0 (best)137.8139.5
ECI rank#99 of 148 (best)#108 of 148#101 of 148
GPQA DiamondGraduate-level science questions75.8%60.8%79.0% (best)
OTIS Mock AIME 2024–2025Competition mathematics88.9% (best)65.3%61.7%
Price per million tokens
Input$0.15$0.07 (best)$0.10
Output$0.60$0.295$0.15 (best)
Cached input———
Blended (3:1)$0.263$0.126$0.113 (best)
Long-context rateSame rateSame rateSame rate
Price sourceMedian of 36 providersMedian of 18 providersMedian of 14 providers
Limits
Context window131,072 tokens131,072 tokens262,144 tokens (best)
Max output32,768 tokens32,768 tokens65,536 tokens (best)
Inputs and features
TextYesYesYes
ImagesNoNoYes
PDFsNoNoNo
AudioNoNoNo
VideoNoNoYes
ReasoningYesYesYes
Tool callingYesYesYes
Structured outputYesYesYes
Availability
WeightsOpenOpenOpen
API model ID———
API providers39 (best)1915
ReleasedAug 5, 2025Aug 5, 2025Feb 23, 2026
Knowledge cutoff———
03 — Cost

What would a month cost?

Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.

  • GPT OSS 120B$2.70
  • GPT OSS 20B$1.29
  • Qwen3.5 9B$1.30
04 — Questions

Which should you choose?

Which is better: GPT OSS 120B, GPT OSS 20B or Qwen3.5 9B?

Qwen3.5 9B is the better all-round choice, scoring 72/100 against GPT OSS 20B (64) and GPT OSS 120B (61). It leads on price, inputs & features and context window. The score weighs capability 50%, price 25%, inputs & features 15%, context window 10%.

Which is cheaper, GPT OSS 120B, GPT OSS 20B or Qwen3.5 9B?

Qwen3.5 9B is cheaper at $0.10 input / $0.15 output per million tokens (median across 14 API providers). GPT OSS 20B costs $0.07 input / $0.295 output per million tokens (median across 18 API providers); GPT OSS 120B costs $0.15 input / $0.60 output per million tokens (median across 36 API providers). At a typical mix of three input tokens to one output token, that is $0.113 per million tokens for Qwen3.5 9B versus $0.126 for GPT OSS 20B (1.1× as much) and $0.263 for GPT OSS 120B (2.3× as much).

Which scores higher on benchmarks?

GPT OSS 120B scores higher on the Capabilities Index (ECI): GPT OSS 120B 140.0 (#99 of 148), Qwen3.5 9B 139.5 (#101 of 148) and GPT OSS 20B 137.8 (#108 of 148). The confidence ranges of the top two overlap (135.3–142.3 vs 136.5–141.3), so treat the gap as small. On individual benchmarks: GPQA Diamond — Qwen3.5 9B 79.0%, GPT OSS 120B 75.8%, GPT OSS 20B 60.8%; OTIS Mock AIME 2024–2025 — GPT OSS 120B 88.9%, GPT OSS 20B 65.3%, Qwen3.5 9B 61.7%.

Which is better for coding?

There are no published SWE-bench Verified results for GPT OSS 120B, GPT OSS 20B and Qwen3.5 9B yet, so there is no like-for-like coding score. On overall capability, GPT OSS 120B leads, which tends to carry over to coding, but test on your own codebase. All three support tool calling for agent workflows.

Which has the bigger context window?

Qwen3.5 9B has the largest context window at 262,144 tokens, against 131,072 for GPT OSS 120B and 131,072 for GPT OSS 20B. Maximum output per response: GPT OSS 120B up to 32,768, GPT OSS 20B up to 32,768, Qwen3.5 9B up to 65,536 tokens.

Which can read images, PDFs, audio or video?

GPT OSS 120B accepts text; GPT OSS 20B accepts text; Qwen3.5 9B accepts text, images and video. Qwen3.5 9B handles the widest range of inputs.

Are any of these open source?

Yes, all three publish their weights, so you can self-host them.

Which is newer?

Qwen3.5 9B is the newest, released Feb 23, 2026. GPT OSS 120B came out Aug 5, 2025; GPT OSS 20B came out Aug 5, 2025.

How do you decide the winner?

Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.

05 — Keep comparing

Related comparisons