Skip to content

fix(deepseek): extract prompt_cache_hit_tokens and reasoning_tokens from usage #1021

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Lintume wants to merge 1 commit into
base: main
Choose a base branch
from
+23 −6
Open

fix(deepseek): extract prompt_cache_hit_tokens and reasoning_tokens from usage #1021

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 8 additions & 2 deletions src/Providers/DeepSeek/Handlers/Stream.php
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -357,9 +357,15 @@ protected function extractUsage(array $data): ?Usage
return null;
}

$totalPrompt = (int) data_get($usage, 'prompt_tokens', 0);
$cacheHit = (int) data_get($usage, 'prompt_cache_hit_tokens', 0);
$reasoning = (int) data_get($usage, 'completion_tokens_details.reasoning_tokens', 0);

return new Usage(
promptTokens: (int) data_get($usage, 'prompt_tokens', 0),
completionTokens: (int) data_get($usage, 'completion_tokens', 0)
promptTokens: max(0, $totalPrompt - $cacheHit),
completionTokens: (int) data_get($usage, 'completion_tokens', 0),
cacheReadInputTokens: $cacheHit > 0 ? $cacheHit : null,
thoughtTokens: $reasoning > 0 ? $reasoning : null,
);
}

Expand Down
10 changes: 8 additions & 2 deletions src/Providers/DeepSeek/Handlers/Text.php
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -122,15 +122,21 @@ protected function sendRequest(Request $request): array
*/
protected function addStep(array $data, Request $request, array $toolResults = []): void
{
$totalPrompt = (int) (data_get($data, 'usage.prompt_tokens') ?? 0);
$cacheHit = (int) (data_get($data, 'usage.prompt_cache_hit_tokens') ?? 0);
$reasoning = (int) (data_get($data, 'usage.completion_tokens_details.reasoning_tokens') ?? 0);

$this->responseBuilder->addStep(new Step(
text: data_get($data, 'choices.0.message.content') ?? '',
finishReason: $this->mapFinishReason($data),
toolCalls: ToolCallMap::map(data_get($data, 'choices.0.message.tool_calls', [])),
toolResults: $toolResults,
providerToolCalls: [],
usage: new Usage(
data_get($data, 'usage.prompt_tokens'),
data_get($data, 'usage.completion_tokens'),
promptTokens: max(0, $totalPrompt - $cacheHit),
completionTokens: (int) (data_get($data, 'usage.completion_tokens') ?? 0),
cacheReadInputTokens: $cacheHit > 0 ? $cacheHit : null,
thoughtTokens: $reasoning > 0 ? $reasoning : null,
),
meta: new Meta(
id: data_get($data, 'id'),
Expand Down
9 changes: 7 additions & 2 deletions tests/Providers/DeepSeek/TextTest.php
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -129,8 +129,13 @@
expect($secondStep->messages[1]->toolCalls[1]->name)->toBe('weather');
expect($secondStep->messages[2])->toBeInstanceOf(ToolResultMessage::class);

// Assert usage
expect($response->usage->promptTokens)->toBe(507);
// Assert usage. promptTokens is now the FRESH portion (prompt_tokens minus
// prompt_cache_hit_tokens) so cost trackers can apply the cached rate to the
// hit portion separately. Aggregated across both steps:
// step 1 fixture: prompt_tokens=220, prompt_cache_hit_tokens=192 → fresh 28, cached 192
// step 2 fixture: prompt_tokens=287, prompt_cache_hit_tokens=256 → fresh 31, cached 256
expect($response->usage->promptTokens)->toBe(59);
expect($response->usage->cacheReadInputTokens)->toBe(448);
expect($response->usage->completionTokens)->toBe(76);

// Assert response
Expand Down