AI & Agent Dev Bug Sandbox Back to Radar

TypeError In _Combine_llm_outputs When Token Usage Contains Nested Dicts

Calling ChatMistralAI.generate() with multiple message lists fails due to unsupported operand type(s) for += on dict in _combine_llm_outputs. The method blindly uses += to aggregate token usage fields, but Mistral API returns nested dictionaries for some fields (e.g., 'tokens_details'), causing a TypeError. A recursive merge is needed.

highConfidence 95%Langchain Affected V1.1.5

Origin Analysis

The _combine_llm_outputs method in langchain_mistralai assumes all token_usage values are scalar numbers and uses += to aggregate them. However, the Mistral API's token usage includes nested dictionary fields, making dict += dict invalid.

1. Install langchain_mistralai version 1.1.5 (or earlier affected versions). 2. Set MISTRAL_API_KEY environment variable. 3. Run the following Python code: ```python import os from langchain_mistralai import ChatMistralAI from langchain_core.messages import HumanMessage llm = ChatMistralAI( model="mistral-small-latest", api_key=os.environ["MISTRAL_API_KEY"], ) result = llm.generate([ [HumanMessage(content="What is 1+1?")], [HumanMessage(content="What is 2+2?")], ]) print(result) ``` 4. Observe TypeError: unsupported operand type(s) for +=: 'dict' and 'dict'.

Fixing Code Block

0x7F 45 4C 46 02 01 01 000x90 12 AB 04 C1 7E 33 08
0xA4 1C 90 3E 77 0B D2 580x11 F6 6A 2C 80 19 DE 44
0x00 00 00 00 03 00 3E 000x01 00 B7 4D 55 90 0C 61
0x6C 69 62 63 2E 73 6F 2E0x36 00 8E 17 42 AA 05 F3
0xDE C0 DE 01 8B 45 FC 890xC7 5D 20 6F 91 03 E4 78
0x13 37 00 4F 9A 2D E1 700x08 B4 5C 22 71 D0 4A 16
0x48 8D 3D 22 01 00 00 E80x91 FF 0A 63 2E 57 B8 04
0x5A 91 C3 0E 44 70 1B 880x02 6F 39 C4 15 80 ED 27
0xFF 15 40 20 60 00 48 890xC3 90 7A 01 4C 28 B1 56
0x0F 1F 44 00 00 55 48 890xE5 C3 18 6D 92 0B 33 A7

Edge Case Audit

The fix assumes that nested dictionaries contain only numeric values or further nested dictionaries. If the Mistral API later returns non-numeric leaf values (e.g., strings or lists), the current merge logic will overwrite instead of summing, potentially losing information. Additionally, this fix only addresses token_usage; other llm_output fields with similar nested structures may still cause errors. Rollback suggestion: if unexpected behavior occurs after applying this patch, revert to using ChatOpenAI with Mistral's OpenAI-compatible endpoint as a workaround, or pin langchain-mistralai to a version known to work.