Coding
Like-for-like- Hy3
- 58.2
- Supported · #28/151
- Inkling
- 41.2
- Supported · #111/151
- Basis
- BenchAlign lane · 0 vs 6 public rows
- Reading
- Hy3 leads
Hy3 vs Inkling
Updated September 10, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
Decision reading
The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. Use the documented cost, context, and runtime rows instead.
0 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Share or export
Which one for your work
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Coding work
Code generation, repair, and software-engineering tasks
Hy3
Hy3 leads on the public coding lane, 58.2 to 41.2, with Supported evidence for both models and non-overlapping 90% intervals.
Confidence: stronger
Long documents
Prompts that approach the documented context limit
Inkling
Inkling has the larger documented context window.
Confidence: documented
Show secondary and unsupported calls
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
Hy3 is scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.
Confidence: limited
Chat turn cost
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Repository review cost
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
What is actually comparable
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
21 Inkling only
- Shared results
- 0
- Hy3 only
- 0
- Inkling only
- 21
- Like-for-like categories
- 1 / 8
2 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Category results, on a stated basis
Each row shows the public-lane category score for both models: the BenchAlign lane for agentic, coding, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.
Agentic
Directional only- Hy3
- 48.4
- Estimated · #67/152
- Inkling
- 38.3
- Supported · #122/152
- Basis
- BenchAlign lane · 0 vs 4 public rows
- Reading
- Directional only
Knowledge
Directional only- Hy3
- 56.4
- Estimated · #48/182
- Inkling
- 57.5
- Supported · #43/182
- Basis
- BenchAlign lane · 0 vs 6 public rows
- Reading
- Directional only
Reasoning
Not comparable- Hy3
- 75.3
- Unranked · 2 rankable rows
- Inkling
- 74.1
- Unranked · 2 rankable rows
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Math
Not comparable- Hy3
- Not ranked
- Inkling
- 78.9
- Unranked · 1 rankable row
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- Hy3
- Not ranked
- Inkling
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
Multimodal
Not comparable- Hy3
- Not ranked
- Inkling
- 49.2
- #38/48
- Basis
- Provisional lane · 0 vs 2 weighted rows
- Reading
- Not comparable
Instruction following
Not comparable- Hy3
- Not ranked
- Inkling
- 85.2
- #43/121
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
| Category | Hy3 | Inkling | Basis | Reading |
|---|---|---|---|---|
| Coding | 58.2Supported · #28/151 | 41.2Supported · #111/151 | Like-for-likeBenchAlign lane · 0 vs 6 public rows | Hy3 leads |
| Agentic | 48.4Estimated · #67/152 | 38.3Supported · #122/152 | Directional onlyBenchAlign lane · 0 vs 4 public rows | Directional only |
| Knowledge | 56.4Estimated · #48/182 | 57.5Supported · #43/182 | Directional onlyBenchAlign lane · 0 vs 6 public rows | Directional only |
| Reasoning | 75.3Unranked · 2 rankable rows | 74.1Unranked · 2 rankable rows | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Math | Not ranked | 78.9Unranked · 1 rankable row | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
| Multimodal | Not ranked | 49.2#38/48 | Not comparableProvisional lane · 0 vs 2 weighted rows | Not comparable |
| Instruction following | Not ranked | 85.2#43/121 | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign) differ from the provisional-lane categories. Unranked scores sit on the lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.
Shape of the matched evidence
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
A shared-evidence shape is not available.
BenchLM does not draw a radar or infer missing axes when the matched evidence is too sparse.
What each workload costs
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
Chat turn
1K fresh input + 500 output tokens
- Hy3
- Self-hosted; infrastructure cost varies
- Fits in one request
- Inkling
- $0.00421
- Fits in one request
Hy3 has no comparable published API token rate.
Repository review
50K fresh input + 3K output tokens
- Hy3
- Self-hosted; infrastructure cost varies
- Fits in one request
- Inkling
- $0.10754
- Fits in one request
Hy3 has no comparable published API token rate.
Cache-heavy agent loop
200K cached + 20K fresh input + 10K output tokens
- Hy3
- Self-hosted; infrastructure cost varies
- Fits in one request
- Cached-input rate unavailable
- Inkling
- $0.159
- Fits in one request
Hy3 has no comparable published API token rate.
Specification differences
Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
SpecificationHy3Inkling
Context window
Maximum documented context; output-token limits may be lower.
Hy3
256K
Inkling
1M
API model ID
Hy3
Not sourced
Inkling
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Hy3
No comparable hosted API rate
Inkling
$0.374 per 1M cached input tokens
Documented inputs
Hy3
Not sourced
Inkling
Not sourced
Documented outputs
Hy3
Not sourced
Inkling
Not sourced
Provider availability
Hy3
Not sourced
Inkling
Not sourced
Reasoning profile
Hy3
Reasoning
Inkling
Hybrid
Weight access
Hy3
Open Weight
Inkling
Open Weight
License
Hy3
Open Weight
Inkling
Open Weight
Release date
Hy3
2026-07-06
Inkling
2026-07-15
If you already use one of these models
- Deployment change
- The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
- Quality signal
- The public evidence has no benchmark result shared by both models, so it does not support a quality verdict.
- Workload cost
- A complete comparable API-rate estimate is not available for both models.
- Context tradeoff
- Inkling has the larger documented window (1M).
Run the same representative tasks against both endpoints before changing production traffic.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence21 rows
Agentic
Coding
SWE-bench Verified
Hy3—Inkling77.6%SourceNot directly comparable
SWE-bench Pro
Hy3—Inkling54.3%SourceNot directly comparable
Terminal-Bench 2.0
Hy3—Inkling63.8%SourceNot directly comparable
FrontierSWE v2
Hy3—Inkling4.1%SourceNot directly comparable
LiveCodeBench (Vals)
Hy3—Inkling85.5%SourceNot directly comparable
SWE-bench (Vals)
Hy3—Inkling77.6%SourceNot directly comparable
Knowledge
GPQA
Hy3—Inkling87.9%SourceNot directly comparable
GPQA-D
Hy3—Inkling87.9%SourceNot directly comparable
HLE
Hy3—Inkling46%SourceNot directly comparable
HLE w/o tools
Hy3—Inkling30%SourceNot directly comparable
GPQA Diamond (Vals)
Hy3—Inkling87.1%SourceNot directly comparable
MMLU-Pro (Vals)
Hy3—Inkling86.3%SourceNot directly comparable
Math
AIME26
Hy3—Inkling97.1%SourceNot directly comparable
Multimodal
Instruction following
IFBench
Hy3—Inkling79.8%SourceNot directly comparable
Frequently asked questions
Which is better, Hy3 or Inkling?
The public evidence has no benchmark result shared by both models, so it does not support a quality verdict. The page therefore keeps the decision tied to the specific documented workload.
Which is better for coding, Hy3 or Inkling?
Hy3 leads the public coding lane, 58.2 to 41.2, with Supported evidence for both models and non-overlapping 90% intervals.
Which is better for agentic tasks, Hy3 or Inkling?
Hy3 scores higher for agentic tasks on the public lane, 48.4 to 38.3. Hy3 is scored on Estimated evidence for agentic tasks, so the reading is directional rather than like-for-like. BenchLM does not name a winner for agentic tasks from a directional reading; read the shared benchmark rows directly and test the models on the same task set.
Which costs less, Hy3 or Inkling?
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Which has the larger context window, Hy3 or Inkling?
Inkling has the larger documented context window: 1M, compared with 256K.
Related comparisons
Last updated September 10, 2026
Watch Hy3 vs Inkling
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.