Source Intelligence
Sweep 28 Aug 2026 · 00:00Z Build v2.1.250 478 read Stable v2.1.236 Latest v2.1.250 Next v2.1.251 Feeds RSS JSON llms.txt
Reading a new release v2.1.251 Analysing changes · 2/5 Deeper second pass · 0/5 agents 427 findings $12.75 so far

DisclaimerUnofficial, and not affiliated with Anthropic. Nearly all of this is read straight out of what ships: npm bundles, captured prompts, published docs. Anthropic's own notes go in verbatim, marked as theirs. The rest is my reading, and every entry carries the strings behind it. If one looks wrong, vote it down and say why.

One change

Retrieve Message Batch results

api/beta/messages/batches/results

first seen The page's own history The capture it came from

Nearest release: v2.1.247, published under an hour before this site recorded the change. Shown because the two are within 24 hours of each other. Nothing here says the release caused the edit.

api/beta/messages/batches/results Changed · +19 / -3 lines

from line 22
 
   - `string`
 
-  - `"message-batches-2024-09-24" or "prompt-caching-2024-07-31" or "computer-use-2024-10-22" or 31 more`
+  - `"message-batches-2024-09-24" or "prompt-caching-2024-07-31" or "computer-use-2024-10-22" or 38 more`
 
     - `"message-batches-2024-09-24"`
 
from line 92
 
     - `"mid-conversation-tool-changes-2026-07-01"`
 
+    - `"compact-2026-01-12"`
+
+    - `"computer-use-2025-11-24"`
+
+    - `"mcp-tunnels-2026-06-22"`
+
+    - `"structured-outputs-2025-11-13"`
+
+    - `"task-budgets-2026-03-13"`
+
+    - `"thinking-display-updates-2026-08-18"`
+
+    - `"ce-user-management-2026-07-13"`
+
 ## Returns
 
 - `BetaMessageBatchIndividualResponse object`
from line 1541
 
             Per-iteration token usage breakdown.
 
-            Each entry represents one sampling iteration, with its own input/output token counts and cache statistics. This allows you to:
+            Each entry represents one sampling iteration, with its own input/output token counts and cache statistics, discriminated by `type`. For `message` entries (model sampling iterations, such as the turns of a server-side tool use loop), this allows you to:
 
             - Determine which iterations exceeded long context thresholds (>=200k tokens)
-            - Calculate the true context window size from the last iteration
+            - Calculate the context window size from the last `message` entry
             - Understand token accumulation across server-side tool use loops
+
+            A `compaction` entry reports the token usage of the compaction operation itself — the server-side request that summarizes the context being closed — NOT the size of the context that was compacted away, and its token counts can be much smaller than that closed context (for example, a compaction that closes a ~200k-token context can report only a few thousand tokens). Do not derive the context window size from a `compaction` entry, even when it is the last entry. A `compaction` entry's tokens are not included in the top-level `usage` fields. When an input-token trigger is in effect (the default — 150,000 tokens unless configured otherwise), each `compaction` entry closes a context that had reached at least that threshold, though the context can exceed it by the final iteration's output and tool results.
 
             - `BetaMessageIterationUsage object`