Quick:
togetherai
Active
MoE
DeepSeek V4 Pro
Model Spec
togetherai:deepseek-ai/DeepSeek-V4-Pro
Modalities
Input
text
Output
text
Capabilities
Chat
Reasoning
Tools
Streaming
Pricing Components
| Component | Price | Conditions | Source |
|---|---|---|---|
| token.input | $1.74 / 1M tokens | Standard | — |
| token.output | $3.48 / 1M tokens | Standard | — |
| token.cache_read | $0.20 / 1M tokens | Standard | — |
Specifications
Architecture
MoE
Total Parameters
1.6T
Active Parameters
49B
Minimum RAM
893.4 GB
Minimum VRAM
819 GB
Context Window
512,000 tokens
Max Output
384,000 tokens
Input Cost
$1.74/M
Output Cost
$3.48/M
History
Raw JSONTimeline is based on collected llm_db history snapshots and may be incomplete.
2026-07-14
changed
-
extra.llmfit.hf_downloads
(replace)
2245489 → 2052463
-
extra.llmfit.hf_likes
(replace)
5021 → 5046
2026-07-02
changed
-
extra.description
(add)
— → Flagship DeepSeek model for coding, reasoning, and agentic work
2026-06-25
changed
-
extra.llmfit.hf_downloads
(replace)
3015772 → 2245489
-
extra.llmfit.hf_likes
(replace)
4963 → 5021
2026-06-24
changed
-
extra.llmfit.hf_downloads
(replace)
4287396 → 3015772
-
extra.llmfit.hf_likes
(replace)
4161 → 4963
2026-06-12
changed
-
cost.input
(replace)
2.1 → 1.74
-
cost.output
(replace)
4.4 → 3.48
-
extra.reasoning_options
(add)
— → [2 items]
2026-05-24
changed
-
extra.llmfit.hf_downloads
(replace)
2967518 → 4287396
-
extra.llmfit.hf_likes
(replace)
3997 → 4161
2026-05-18
changed
-
extra.llmfit.hf_downloads
(replace)
1167697 → 2967518
-
extra.llmfit.hf_likes
(replace)
3780 → 3997
2026-05-11
changed
-
extra.llmfit.hf_downloads
(replace)
786631 → 1167697
-
extra.llmfit.hf_likes
(replace)
3602 → 3780
2026-05-07
changed
-
extra.llmfit.hf_downloads
(replace)
174402 → 786631
-
extra.llmfit.hf_likes
(replace)
3169 → 3602
2026-04-30
changed
-
extra.llmfit.hf_downloads
(replace)
30 → 174402
-
extra.llmfit.hf_likes
(replace)
2180 → 3169
2026-04-26
introduced
snapshots 64bf744 -> b7ca903 • generated 2026-08-05T03:31:59
See incorrect model data? Help us improve it.
Submit Fix on GitHub
6293
of 6427 models
1 / 126
Page 1 of 126