DeepSeek v4 and Claude pricing data gaps

NousResearch/hermes-agent · cluster 24141 · agent

Data as of 2026-10-22 12:15 UTC

2 open items. Fix merged in #24579. Check it covers your case.

open issues

open PRs

items grouped

Known?: Yes. 14 related items grouped here, 2 issues already closed.
Fix event: A fix was merged to main in PR #24579. Check it covers your case. No release data is tracked.
Tracker: The maintainers' tracker marks this partially resolved, deciding PR #65607. That is separate from the fix event.
Still open: 2 issues and 0 PRs.

Next step: read #24579, then check whether your report is one of the 2 still open.

Read #24579 Pick up a Lab task

Most members concern missing or incorrect pricing entries for DeepSeek v4-pro and v4-flash models, including a reported 4x overcharge on v4-pro rates. A smaller subset addresses a missing Claude Fable 5 pricing entry with similar unknown-cost symptoms.

Open issues2

Issue#24141insights: DeepSeek v4-pro/v4-flash pricing not matching + cost not rendered in output
Issue#43321claude-fable-5 missing from _OFFICIAL_DOCS_PRICING → $0/unknown cost on direct Anthropic

Open PRs0

No open pull requests.

Fixes on record

Merged to main. A stamp counts the issues a fix is linked to.

#24579 fix(pricing): add deepseek-v4-pro to official docs pricing table

merged to main

files: agent/usage_pricing.py, scripts/release.py, tests/agent/test_usage_pricing.py

#65607 fix(pricing): add deepseek-v4-flash entry + refresh stale DeepSeek snapshot to 2026-07 rates

merged to main

files: agent/usage_pricing.py, scripts/release.py, tests/agent/test_usage_pricing.py

Show the diff

agent/usage_pricing.py

@@ -446,36 +446,52 @@ class CostResult:
         pricing_version="anthropic-pricing-2026-05",
     ),
     # DeepSeek
+    # Snapshot of https://api-docs.deepseek.com/quick_start/pricing (2026-07).
+    # deepseek-chat / deepseek-reasoner are deprecated 2026-07-24 and now alias
+    # deepseek-v4-flash's non-thinking / thinking modes — same rates.
     (
         "deepseek",
         "deepseek-chat",
     ): PricingEntry(
         input_cost_per_million=Decimal("0.14"),
         output_cost_per_million=Decimal("0.28"),
+        cache_read_cost_per_million=Decimal("0.0028"),
         source="official_docs_snapshot",
         source_url="https://api-docs.deepseek.com/quick_start/pricing",
-        pricing_version="deepseek-pricing-2026-03-16",
+        pricing_version="deepseek-pricing-2026-07",
     ),
     (
         "deepseek",
         "deepseek-reasoner",
     ): PricingEntry(
-        input_cost_per_million=Decimal("0.55"),
-        output_cost_per_million=Decimal("2.19"),
+        input_cost_per_million=Decimal("0.14"),
+        output_cost_per_million=Decimal("0.28"),
+        cache_read_cost_per_million=Decimal("0.0028"),
         source="official_docs_snapshot",
         source_url="https://api-docs.deepseek.com/quick_start/pricing",
-        pricing_version="deepseek-pricing-2026-03-16",
+        pricing_version="deepseek-pricing-2026-07",
     ),
     (
         "deepseek",
         "deepseek-v4-pro",
     ): PricingEntry(
-        input_cost_per_million=Decimal("1.74"),
-        output_cost_per_million=Decimal("3.48"),
-        cache_read_cost_per_million=Decimal("0.0145"),
+        input_cost_per_million=Decimal("0.435"),
+        output_cost_per_million=Decimal("0.87"),
+        cache_read_cost_per_million=Decimal("0.003625"),
+        source="official_docs_snapshot",
+        source_url="https://api-docs.deepseek.com/quick_start/pricing",
+        pricing_version="deepseek-pricing-2026-07",
+    ),
+    (
+        "deepseek",
+        "deepseek-v4-flash",
+    ): PricingEntry(
+        input_cost_per_million=Decimal("0.14"),
+        output_cost_per_million=Decimal("0.28"),
+        cache_read_cost_per_million=Decimal("0.0028"),
         source="official_docs_snapshot",
         source_url="https://api-docs.deepseek.com/quick_start/pricing",
-        pricing_version="deepseek-pricing-2026-05-12",
+        pricing_version="deepseek-pricing-2026-07",
     ),
     # Google Gemini
     (

scripts/release.py

@@ -351,6 +351,7 @@
     "[redacted:email]": "dirtyren",
     "[redacted:email]": "Code-suphub",
     "[redacted:email]": "Code-suphub",
+    "[redacted:email]": "shuangxinniao",
     "[redacted:email]": "brendandebeasi",
     "[redacted:email]": "dorokuma",
     "[redacted:email]": "liuwei666888",

tests/agent/test_usage_pricing.py

@@ -223,7 +223,8 @@ def test_deepseek_v4_pro_pricing_entry_exists():
 
     Before this fix, deepseek-v4-pro sessions showed as unknown cost
     in hermes insights because the _OFFICIAL_DOCS_PRICING table had no
-    entry for that model.  See #24218.
+    entry for that model.  See #24218.  Rates track the 2026-07 price cut
+    ($1.74/$3.48 → $0.435/$0.87).
     """
     entry = get_pricing_entry(
         "deepseek-v4-pro",
@@ -233,9 +234,9 @@ def test_deepseek_v4_pro_pricing_entry_exists():
     assert entry is not None
     assert entry.input_cost_per_million is not None
     assert entry.output_cost_per_million is not None
-    assert float(entry.input_cost_per_million) == 1.74
-    assert float(entry.output_cost_per_million) == 3.48
-    assert float(entry.cache_read_cost_per_million) == 0.0145
+    assert float(entry.input_cost_per_million) == 0.435
+    assert float(entry.output_cost_per_million) == 0.87
+    assert float(entry.cache_read_cost_per_million) == 0.003625
 
 
 def test_deepseek_v4_pro_estimate_usage_cost():
@@ -248,8 +249,39 @@ def test_deepseek_v4_pro_estimate_usage_cost():
 
     assert result.status == "estimated"
     assert result.amount_usd is not None
-    # 1M input × $1.74/M + 500K output × $3.48/M = $1.74 + $1.74 = $3.48
-    assert float(result.amount_usd) == 3.48
+    # 1M input × $0.435/M + 500K output × $0.87/M = $0.435 + $0.435 = $0.87
+    assert float(result.amount_usd) == 0.87
+
+
+def test_deepseek_deprecated_aliases_price_as_v4_flash():
+    """Invariant: deepseek-chat / deepseek-reasoner are deprecated aliases for
+    deepseek-v4-flash's non-thinking / thinking modes (deprecation 2026-07-24)
+    — they must bill at identical rates to the flash entry, or sessions on the
+    legacy names over/under-report cost."""
+    flash = get_pricing_entry("deepseek-v4-flash", provider="deepseek")
+    assert flash is not None
+    for alias in ("deepseek-chat", "deepseek-reasoner"):
+        entry = get_pricing_entry(alias, provider="deepseek")
+        assert entry is not None, alias
+        assert entry.input_cost_per_million == flash.input_cost_per_million, alias
+        assert entry.output_cost_per_million == flash.output_cost_per_million, alias
+        assert (
+            entry.cache_read_cost_per_million == flash.cache_read_cost_per_million
+        ), alias
+
+
+def test_deepseek_rows_all_carry_cache_read_pricing():
+    """Invariant: DeepSeek publishes a cache-hit rate for every current model;
+    every deepseek snapshot row must carry cache_read < input so cached
+    sessions estimate correctly instead of billing reads at full price."""
+    from agent.usage_pricing import _OFFICIAL_DOCS_PRICING
+
+    ds_rows = [k for k in _OFFICIAL_DOCS_PRICING if k[0] == "deepseek"]
+    assert ds_rows, "expected at least one deepseek pricing row"
+    for key in ds_rows:
+        entry = _OFFICIAL_DOCS_PRICING[key]
+        assert entry.cache_read_cost_per_million is not None, key
+        assert entry.cache_read_cost_per_million < entry.input_cost_per_million, key
 
 
 def test_bedrock_claude_rows_all_carry_cache_pricing():
@@ -415,3 +447,37 @@ def test_fireworks_rows_all_carry_cache_read_pricing():
         entry = _OFFICIAL_DOCS_PRICING[key]
         assert entry.cache_read_cost_per_million is not None, key
         assert entry.cache_read_cost_per_million < entry.input_cost_per_million, key
+
+
+def test_deepseek_v4_flash_pricing_entry_exists():
+    """Regression test: deepseek-v4-flash must have a pricing entry.
+
+    Before this fix, deepseek-v4-flash sessions showed $0.00 / cost_source
+    "none" because the _OFFICIAL_DOCS_PRICING table had an entry for
+    deepseek-v4-pro but not the (newer) flash model.  DeepSeek's /models
+    endpoint returns no pricing, so the official-docs snapshot is the only
+    source for direct-provider routes.
+    """
+    entry = get_pricing_entry(
+        "deepseek-v4-flash",
+        provider="deepseek",
+    )
+
+    assert entry is not None
+    assert float(entry.input_cost_per_million) == 0.14
+    assert float(entry.output_cost_per_million) == 0.28
+    assert float(entry.cache_read_cost_per_million) == 0.0028
+
+
+def test_deepseek_v4_flash_estimate_usage_cost():
+    """Ensure deepseek-v4-flash sessions get a dollar estimate, not $0/none."""
+    result = estimate_usage_cost(
+        "deepseek-v4-flash",
+        CanonicalUsage(input_tokens=[redacted:secret-assignment] output_tokens=[redacted:secret-assignment]
+        provider="deepseek",
+    )
+
+    assert result.status == "estimated"
+    assert result.amount_usd is not None
+    # 1M input × $0.14/M + 500K output × $0.28/M = $0.14 + $0.14 = $0.28
+    assert float(result.amount_usd) == 0.28

8 PRs closed without merging

From the close comments: merged via another change 3 · closed without comment 1 · done elsewhere 1 · duplicate 1 · superseded by another PR 1 · other 1

#24195 fix(insights): show DeepSeek costs · closed without comment
#24216 feat(pricing): add deepseek-v4-pro to docs pricing snapshot · superseded by another PR → #24579, #24327
#24327 fix: add pricing for deepseek-v4-pro model · merged via another change → #24579
#32035 fix(pricing): add deepseek-v4-flash to official docs pricing table · merged via another change → #40127, #41719, #65607
#41719 feat(pricing): add deepseek-v4-flash pricing entry · merged via another change → #40127, #65607, #41710

Show 3 more

#41723 fix(pricing): refresh stale deepseek-v4-pro rates (4x overstated) · done elsewhere → #65607, #41719
#43317 fix(pricing): add Claude Fable 5 usage pricing · other
#59095 [codex] refresh DeepSeek v4 pricing · duplicate → #32035, #41723, #41719, #40127, #41710

Closed issues · 2

Closed#24218[Bug]: deepseek-v4-pro sessions show as unknown cost in `hermes insights`
Closed#41710deepseek-v4-flash missing from _OFFICIAL_DOCS_PRICING → $0/unknown cost

How this cluster was grouped

Link evidence (partly structural links): some members are connected by structural evidence (shared changed files, or a closing/superseded reference); the rest are joined by an automated relevance judgement over their text.
Members outside that structural core may belong to a different underlying problem.
Pairwise link density 0.14: the share of item pairs that carry a link, counting a structural link in full and a relevance-only link at 0.36. 1.00 means every pair is linked.

Pick up a Lab task

Check whether a fix covers the open issues here, with your own agent. Bring back evidence for maintainers.

Open the Lab

At a glance

Subsystem: agent
Merged PRs: #24579, #65607
Closed issues: 2
Attempts: 8 PRs closed without merging
Fix link checked: 16 Jul 2026

For your agent

The same facts, typed.

cluster/24141.json
cluster/24141.md