Baseten alternatives: API cost and speed

Baseten, Novita AI and Together AI charge the same $0.0042 for the DeepSeek V4.1 Flash 1017 example and $0.0184 for GLM-5.3. Endpoint performance distinguishes these equal uncached tariffs; GPT-OSS 120B remains a separate baseline.

Updated · 11 sources · Methodology

NeedPickWhy
Lowest GPT-OSS alternative costNovita AIGPT-OSS 120B · $0.000752
Fastest GPT-OSS alternative outputCerebras1766.7 tokens/s
Shortest GPT-OSS alternative initial waitCerebras0.38 s
Shortest DeepSeek initial waitTogether AIDeepSeek V4.1 Flash 1017 · Together AI 0.30 s / Novita AI 1.94 s7
GLM alternative output speedTogether AIGLM-5.3 · Together AI 152.6 tokens/s / Novita AI 118.3 tokens/s9

Performance

GPT-OSS 120B · Baseline

ProviderOutput speed (tokens/s)First-chunk median (s)Endpoint Accuracy Index v1.0
Novita AI1236.41.0191%
Groq1469.90.6686%
Together AI1100.80.4877%
Cerebras11766.70.3887%

DeepSeek V4.1 Flash 1017 (Max)

ProviderOutput speed (tokens/s)AA median first chunk (s)
Baseten7247.70.31
Novita AI7167.21.94
Together AI7349.30.3

GLM-5.3 (Max)

ProviderOutput speed (tokens/s)AA median first chunk (s)
Baseten9156.20.91
Novita AI9118.32.30
Together AI9152.60.65

Prices and request cost

GPT-OSS 120B · Baseline

ProviderInput / 1M tokensOutput / 1M tokens
Baseten3$0.1$0.5
Novita AI2$0.05$0.25
Groq4$0.15$0.6
Together AI5$0.15$0.6
Cerebras6$0.35$0.75
API documentationNovita AI · Chat API
Cost example: 10,000 input + 1,000 billed output tokens
Baseten$0.0015
Novita AI$0.00075
Groq$0.0021
Together AI$0.0021
Cerebras$0.00425

DeepSeek V4.1 Flash 1017

ProviderInput / 1M tokensOutput / 1M tokens
Baseten3$0.3$1.2
Novita AI10$0.3$1.2
Together AI8$0.3$1.2
API documentationNovita AI · Chat API
DeepSeek V4.1 Flash 1017 · 10,000 input + 1,000 output tokens
Baseten$0.0042
Novita AI$0.0042
Together AI$0.0042

GLM-5.3

ProviderInput / 1M tokensOutput / 1M tokens
Baseten3$1.4$4.4
Novita AI11$1.4$4.4
Together AI8$1.4$4.4
API documentationNovita AI · Chat API
GLM-5.3 · 10,000 input + 1,000 output tokens
Baseten$0.0184
Novita AI$0.0184
Together AI$0.0184

Novita endpoint limits

ModelContext window (tokens)Max output (tokens)
GPT-OSS 120B2131,07232,768
DeepSeek V4.1 Flash 1017101,048,576393,216
GLM-5.3111,048,576131,072
API documentationNovita AI · Chat API

Options, strengths and limitations

  1. Novita AI

    For GPT-OSS 120B, Novita AI combines 236.4 tokens/s with the shortlist’s lowest example cost.1

    Input / 1M tokens
    $0.05
    Output / 1M tokens
    $0.25
    Output speed (tokens/s)
    236.4

    Strengths

    • GPT-OSS 120B has a $0.00075 illustrated request cost on Novita AI; the measured initial wait is 1.01 s.21

    Limitations

    • At 236.4 tokens/s, Novita trails Cerebras at 1766.7 tokens/s.1

    Sources:21

  2. Groq

    Groq’s GPT-OSS 120B endpoint delivers 469.9 tokens/s, compared with Novita AI’s 236.4 tokens/s.1

    Input / 1M tokens
    $0.15
    Output / 1M tokens
    $0.6
    Output speed (tokens/s)
    469.9

    Strengths

    • A 0.66 s initial wait accompanies the $0.0021 GPT-OSS 120B request example.41

    Limitations

    • The request example costs $0.0021, versus $0.00075 on Novita AI.41

    Sources:41

  3. Together AI

    Together AI offers a direct-account GPT-OSS 120B alternative at 100.8 tokens/s.1

    Input / 1M tokens
    $0.15
    Output / 1M tokens
    $0.6
    Output speed (tokens/s)
    100.8

    Strengths

    • Together AI combines a $0.0021 request example for the GPT-OSS 120B token scenario with a 0.48 s initial wait.51

    Limitations

    • Together AI’s 100.8 tokens/s remains below Novita AI’s 236.4 tokens/s in this table.51

    Sources:51

  4. Cerebras

    Cerebras leads the displayed GPT-OSS 120B output table at 1766.7 tokens/s.1

    Input / 1M tokens
    $0.35
    Output / 1M tokens
    $0.75
    Output speed (tokens/s)
    1766.7

    Strengths

    • The GPT-OSS 120B baseline shows a 0.38 s first-chunk wait; the illustrated request is $0.00425.61

    Limitations

    • The request example costs $0.00425, versus $0.00075 on Novita AI.61

    Sources:61

Baseten’s original endpoint

For GPT-OSS 120B at 10,000 input tokens, AA lists Baseten at 178.8 tokens/s and a 0.19 s median first chunk. AA reports a 98% Endpoint Accuracy Index for this baseline.1

Methodology and sources

For GPT-OSS 120B, AA performance uses 10,000 input tokens and high reasoning: output speed and first reasoning-chunk latency are medians. Endpoint Accuracy Index comes from a separate AA evaluation relative to a reference endpoint; it is not a pass rate. The uncached cost example bills 10,000 input and 1,000 output tokens at the named sellers’ standard tariffs. The DeepSeek V4.1 Flash 1017 and GLM-5.3 groups compare the listed Max configurations within each model. AA derives their first-answer latency estimates from initial response and reasoning duration, not a separately sampled median. Novita AI and Together AI are the current-model alternatives; Baseten is additionally included on its own page. GPT-OSS 120B remains the common baseline for the other services, not a ranking of every model they offer. Rolling third-party performance values are the batch snapshot identified by the page’s Updated line, not live guarantees.

  1. Artificial Analysis · GPT-OSS 120B provider table
  2. Novita AI · GPT-OSS 120B endpoint
  3. Baseten · API pricing
  4. Groq · API pricing
  5. Together AI · API pricing
  6. Cerebras · API pricing
  7. Artificial Analysis · DeepSeek V4.1 Flash 1017 provider comparison
  8. Together AI · official model pricing
  9. Artificial Analysis · GLM-5.3 provider comparison
  10. Novita AI · DeepSeek V4.1 Flash 1017 endpoint
  11. Novita AI · GLM-5.3 endpoint

Build with Novita Model APIs

Create an API key and connect your application.

Start BuildingExplore Models