Skip to content

Commit 8d68939

Browse files
committed
fix: update Deepseek configurations
1 parent 5971caa commit 8d68939

1 file changed

Lines changed: 13 additions & 37 deletions

File tree

packages/types/src/providers/deepseek.ts

Lines changed: 13 additions & 37 deletions
Original file line numberDiff line numberDiff line change
@@ -14,56 +14,32 @@ export const deepSeekModels = {
1414
contextWindow: 1_000_000,
1515
supportsImages: true,
1616
supportsPromptCache: true,
17-
supportsReasoningEffort: ["disable", "low", "medium", "high", "xhigh"],
17+
supportsReasoningEffort: ["low", "high", "max"], // Updated 2026-08-01
1818
preserveReasoning: true,
1919
reasoningEffort: "high",
20-
inputPrice: 0.14, // $0.14 per million tokens (cache miss) - Updated Apr 29, 2026
21-
outputPrice: 0.28, // $0.28 per million tokens - Updated Apr 29, 2026
22-
cacheWritesPrice: 0.14, // $0.14 per million tokens (cache miss) - Updated Apr 29, 2026
23-
cacheReadsPrice: 0.0028, // $0.0028 per million tokens (cache hit) - Updated Apr 29, 2026
20+
inputPrice: 0, // the inputs are priced as cache read/write, so `inputPrice` should be 0
21+
// the peak/off-peak pricing policy has not been implemented yet - Updated 2026-08-01
22+
outputPrice: 0.28, // $0.28 per million tokens - Updated 2026-08-01
23+
cacheWritesPrice: 0.14, // $0.14 per million tokens (cache miss) - Updated 2026-08-01
24+
cacheReadsPrice: 0.0028, // $0.0028 per million tokens (cache hit) - Updated 2026-08-01
2425
description: `DeepSeek-V4-Flash is DeepSeek's fast, cost-efficient V4 model. It supports thinking and non-thinking modes, JSON output, tool calls, chat prefix completion (beta), and FIM completion (beta) in non-thinking mode.`,
2526
},
2627
"deepseek-v4-pro": {
2728
maxTokens: 384_000,
2829
contextWindow: 1_000_000,
2930
supportsImages: true,
3031
supportsPromptCache: true,
31-
supportsReasoningEffort: ["disable", "low", "medium", "high", "xhigh"],
32+
supportsReasoningEffort: ["high", "max"], // Updated 2026-08-01
3233
preserveReasoning: true,
3334
reasoningEffort: "high",
34-
// TODO(deepseek): Re-check V4 Pro discounted prices after DeepSeek's 2026-05-31 discount end date.
35-
inputPrice: 0.435, // $0.435 per million tokens (cache miss, discounted) - Updated Apr 29, 2026
36-
outputPrice: 0.87, // $0.87 per million tokens (discounted) - Updated Apr 29, 2026
37-
cacheWritesPrice: 0.435, // $0.435 per million tokens (cache miss, discounted) - Updated Apr 29, 2026
38-
cacheReadsPrice: 0.003625, // $0.003625 per million tokens (cache hit, discounted) - Updated Apr 29, 2026
35+
inputPrice: 0, // the inputs are priced as cache read/write, so `inputPrice` should be 0
36+
// the peak/off-peak pricing policy has not been implemented yet - Updated 2026-08-01
37+
outputPrice: 0.87, // $0.87 per million tokens - Updated 2026-08-01
38+
cacheWritesPrice: 0.435, // $0.435 per million tokens (cache miss) - Updated 2026-08-01
39+
cacheReadsPrice: 0.003625, // $0.003625 per million tokens (cache hit) - Updated 2026-08-01
3940
description: `DeepSeek-V4-Pro is DeepSeek's strongest V4 model for reasoning, coding, long-context, and agentic workloads. It supports thinking and non-thinking modes, JSON output, tool calls, chat prefix completion (beta), and FIM completion (beta) in non-thinking mode.`,
4041
},
41-
// TODO(deepseek): Remove this compatibility alias after DeepSeek's 2026-07-24 retirement date.
42-
"deepseek-chat": {
43-
maxTokens: 8192, // 8K max output
44-
contextWindow: 128_000,
45-
supportsImages: false,
46-
supportsPromptCache: true,
47-
inputPrice: 0.28, // $0.28 per million tokens (cache miss) - Updated Dec 9, 2025
48-
outputPrice: 0.42, // $0.42 per million tokens - Updated Dec 9, 2025
49-
cacheWritesPrice: 0.28, // $0.28 per million tokens (cache miss) - Updated Dec 9, 2025
50-
cacheReadsPrice: 0.028, // $0.028 per million tokens (cache hit) - Updated Dec 9, 2025
51-
description: `Legacy compatibility alias for the non-thinking mode of deepseek-v4-flash. DeepSeek plans to deprecate this model name on 2026-07-24.`,
52-
},
53-
// TODO(deepseek): Remove this compatibility alias after DeepSeek's 2026-07-24 retirement date.
54-
"deepseek-reasoner": {
55-
maxTokens: 8192, // 8K max output
56-
contextWindow: 128_000,
57-
supportsImages: false,
58-
supportsPromptCache: true,
59-
preserveReasoning: true,
60-
inputPrice: 0.28, // $0.28 per million tokens (cache miss) - Updated Dec 9, 2025
61-
outputPrice: 0.42, // $0.42 per million tokens - Updated Dec 9, 2025
62-
cacheWritesPrice: 0.28, // $0.28 per million tokens (cache miss) - Updated Dec 9, 2025
63-
cacheReadsPrice: 0.028, // $0.028 per million tokens (cache hit) - Updated Dec 9, 2025
64-
description: `Legacy compatibility alias for the thinking mode of deepseek-v4-flash. DeepSeek plans to deprecate this model name on 2026-07-24.`,
65-
},
6642
} as const satisfies Record<string, ModelInfo>
6743

6844
// https://api-docs.deepseek.com/quick_start/parameter_settings
69-
export const DEEP_SEEK_DEFAULT_TEMPERATURE = 0.3
45+
export const DEEP_SEEK_DEFAULT_TEMPERATURE = 0.0

0 commit comments

Comments
 (0)