DeepSeek on Svtter's Blog

DeepSeek + Claude Code: Thinking Block Compatibility Analysis

Thu, 30 Apr 2026 15:00:00 +0800

Problem Description

When using DeepSeek models (such as deepseek-v4-flash) directly in Claude Code with extended thinking enabled, multi-turn conversations trigger a 400 error:

1

Bad Request: {"error":{"message":"The content[].thinking in the thinking mode must be passed back to the API.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}

Root Cause Analysis

Call Chain

1

Claude Code → DeepSeek Anthropic Compatible Endpoint (https://api.deepseek.com/anthropic)

Protocol Incompatibility

According to the DeepSeek Anthropic API Compatibility Documentation, the compatibility status is as follows:

Message Field	Support Status
`content[].thinking`	✅ Supported
`content[].redacted_thinking`	❌ Not Supported

In extended thinking mode during multi-turn conversations, Claude Code faithfully passes back all thinking blocks from the previous round (including redacted_thinking types) to the API as-is. DeepSeek does not recognize redacted_thinking, hence the 400 error.

Additionally, DeepSeek’s thinking block format differs from Anthropic’s native protocol, and the replay logic in tool_use scenarios is not fully compatible either.

Core Conflict

Anthropic API requirement: In extended thinking mode, content[].thinking and content[].redacted_thinking must be passed back unchanged
DeepSeek compatibility layer: Only supports thinking, does not support redacted_thinking
Claude Code behavior: Hard-coded according to Anthropic protocol, does not distinguish between target endpoint types

Community Feedback

This is a widespread community issue that almost all CC agent/router projects have encountered:

Issue	Project	Title
#1	cc-use	DeepSeek Thinking Mode Error: `content[].thinking` Must Be Passed Back
#878	openclaude	DeepSeek V4: reasoning_content must be passed back (400) on tool_calls
#1355	claude-code-router	CCR 代理 deepseek V4 思考时返回 400
#4543	new-api	ClaudeCode 接入 DeepSeek V4 遇到 400 reasoning_content 报错
#355	9router	DeepSeek API Error 400 – Missing reasoning_content
#16748	hermes-agent	DeepSeek /anthropic: stripped thinking blocks cause HTTP 400 on replay
#2414	cc-switch	Claude 使用 cc-switch 配置 deepseek-v4-pro，无法识别字段
#174	cc-haha	/compact 命令在使用 DeepSeek API 时无法工作

DeepSeek Official Response

Zero response. Nor is there any need to respond.

First, DeepSeek has no public API issue repository. All feedback occurs in third-party projects without any DeepSeek official personnel participating in any discussions.
Second, whether to use Anthropic as a compatibility standard, I think DeepSeek should be hesitant.

Temporary Workarounds

Disable extended thinking — When using DeepSeek in CC, turn off thinking mode
Use proxy filtering — Add a proxy layer between CC and DeepSeek to filter out redacted_thinking blocks
Switch models — Use DeepSeek for non-thinking scenarios and Anthropic native models for thinking scenarios

Why Doesn’t OpenCode Have This Problem?

OpenCode (opencode-ai/opencode) naturally avoids this problem architecturally, not through a dedicated “fix”.

The key lies in the convertMessages method in internal/llm/provider/anthropic.go (lines 60-119):

When building assistant messages, it only passes back TextContent (text) and ToolCall (tool calls)
Completely ignores ReasoningContent (thinking content), not putting it in messages
thinking content is only displayed in the UI through stream thinking_delta events and is not passed back to the API

Comparison with Claude Code’s behavior:

	Claude Code	OpenCode
thinking replay	✅ Faithfully replay all thinking blocks (including redacted_thinking)	❌ Do not replay thinking blocks
architectural reason	Follow Anthropic API specification, requires unchanged replay	Self-managed conversation state, thinking only for UI display
DeepSeek compatibility	❌ Triggers 400 (redacted_thinking not recognized)	✅ Not affected (doesn’t pass thinking at all)

Conclusion: OpenCode avoids the problem at the cost of not following Anthropic’s extended thinking specification. This approach is friendly to third-party compatible endpoints like DeepSeek, but if Anthropic native thinking context retention capability is needed in the future, re-implementation may be necessary.

Does Not Replay Thinking Blocks Affect DeepSeek Performance?

Basically no, reasons:

thinking blocks are the model’s internal scratchpad, not final output. The text replies and tool calls in the conversation history already retain key decisions and conclusions
DeepSeek’s reasoning is closer to OpenAI’s mode — each round is generated independently, unlike Anthropic’s strong reliance on cross-round replay to maintain reasoning coherence
OpenCode’s extensive actual use also confirms this — community users run multi-turn conversations using DeepSeek thinking mode in OpenCode without feedback about reasoning quality degradation

The truly potentially affected extreme scenario: in ultra-long multi-turn tasks, the model may repeat conclusions it has already reasoned through. However, in most actual use, the impact is negligible.

CC itself has similar thinking block replay bugs on Anthropic models (not DeepSeek-specific):

Issue	Title	Status
#10199	API Error 400 - Thinking Block Modification Error	Open (oncall)
#51985	thinking block missing in multi-turn conversations	Open
#20692	thinking blocks order error on first tool use	Open (oncall)
#54482	Thinking blocks stripped from context every turn (Opus 4.7)	Open

How to Fix DeepSeek Model Reasoning Issues in OpenCode

Fri, 24 Apr 2026 12:23:58 +0800

When using deepseek-reasoner, we often encounter this problem:

1

The reasoning_content' in the thinking mode must be passed back to the API.

Update

Both issues have now been officially resolved by opencode. Users only need to install the latest version of opencode and use it through the deepseek provider, without additional configuration.

1
2
3
4
5
6


Issue 1
The reasoning_content' in the thinking mode must be passed back to the API.

Issue 2
Bad Request: {"error":{"message":"The content[].thinking in the thinking mode must be passed back to the
API.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}

Both issues have been officially resolved. Install version 1.14.29 or above.

The old solution follows:

How to solve it? It’s straightforward.

How to Configure

Add provider information to your configuration:

.config/opencode/opencode.json or .config/opencode/opencode.jsonc

Modify the provider section to:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40


{
 "provider": {
 "deepseek": {
 "npm": "@ai-sdk/anthropic",
 "name": "DeepSeek",
 "options": {
 "baseURL": "https://api.deepseek.com/anthropic",
 "apiKey": "<apikey>"
 },
 "models": {
 "deepseek-v4-pro": {
 "name": "DeepSeek-V4-Pro",
 "limit": {
 "context": 1048576,
 "output": 262144
 },
 "options": {
 "thinking": {
 "type": "enabled",
 "budgetTokens": 8192
 }
 }
 },
 "deepseek-v4-flash": {
 "name": "DeepSeek-V4-Flash",
 "limit": {
 "context": 1048576,
 "output": 262144
 },
 "options": {
 "thinking": {
 "type": "enabled",
 "budgetTokens": 8192
 }
 }
 }
 }
 }
 }
}

How to Use

Select the deepseek model.

The result.

Supplement

This method cannot solve this problem

Bad Request: {"error":{"message":"The content[].thinking in the thinking mode must be passed back to the API.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}

If you encounter this problem, you need to wait for opencode to fix it.

Related article: DeepSeek + Claude Code: Thinking Block Compatibility Issue Analysis — Analyzes the root cause of 400 errors triggered by multi-turn conversations in extended thinking mode when using DeepSeek with Claude Code, along with community solutions.

Claude Code Plugin Usage Experience

Tue, 14 Oct 2025 10:16:54 +0800

Overall, the experience was not good.

It’s likely because it’s newly launched and generally feels immature.

Typical issues include:

Not using available agents.
Not using available MCP.

Tool calls are infrequent and require manual prompting. As a user, I generally don’t deliberately memorize which agents are available.

More importantly, it impacts efficiency.

If using DeepSeek V3.2, its relatively short context length (128K) means it doesn’t perform well when there are many tools or MCP connections.
Plugins often don’t improve the tool usage experience; they can actually degrade it. This is because MCP tools and plugins increase the input token count, forcing the model to process more context. Since the computational complexity of transformers is O(n²), any increase in length has a significant negative impact.

In summary, it’s not recommended for use at this time.

[Expired] I now use GLM 4.6 more often.

Thu, 09 Oct 2025 15:36:00 +0800

1
2
3


● Update(content/post/2025-10-24-我又买了-kimi-coding-plan/pics/bg.svg)
 ⎿ Error editing file
 ⎿ Interrupted · What should Claude do instead?

updated at: 2025-10-27 I only use glm4.6 for very simple tasks. In practical experience, minor issues frequently arise. For example, when using claude code, it is unable to update files. Here are some recent experiences using code agents.

Model Comparison

Based on my practical usage, GLM 4.6 is still slightly stronger than DeepSeek v3.2.

For example, in a Next.js project, I configured nextjs config -> baseUrl 192.168.2.14:8080. GLM 4.6 was able to recognize this pre-configured setting without explicit context, whereas DeepSeek v3.2 could not.

However, GLM 4.6 is not superior in all aspects. When dealing with relatively ambiguous problems, DeepSeek v3.2 is more conservative and does not violate the constraints I set before task completion. In contrast, GLM 4.6 tends to ignore my constraints, makes bold modifications, and ends up breaking things.

Tools

Compared to using GLM 4.6 in Claude Code / Cline, the experience in Kilo Code is the best.

Kilo Code can read files in parallel, while CC can only read them one by one.
Kilo Code enforces the generation of a plan, imposing more restrictions on the big model compared to CC.
The visual interface is more user-friendly. I can directly ban Python commands (I need to execute uv run instead of directly running Python commands).

However, Kilo Code itself also has issues. It cannot use MCP servers of the input; http type, which prevents the use of web-search-prime on Kilo Code.

How to Use Claude Code With Deepseek

Tue, 26 Aug 2025 14:42:54 +0800

Sometimes we cannot directly use the Anthropic API. However, the Claude Code (CC) experience is excellent, and we still want to use CC.

In such cases, you can try using the API provided by DeepSeek to access CC.

DeepSeek has already provided the corresponding interface: How to Use the Claude Code + DeepSeek Combination?

Currently, there are two mainstream LLM APIs: one is OpenAI, and the other is Anthropic. Anthropic has gained a certain level of influence through CC.

If you want to learn more about the use cases for CC, I recommend reading Anthropic’s Official Blog.

Additionally, here are some supplementary resources:

https://mp.weixin.qq.com/s/gk0tzMxWZ-NgsUWg5iLoSg